diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,14421 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.5677651344893663,
+  "eval_steps": 500,
+  "global_step": 12000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999998619058638e-06,
+      "loss": 7.2737,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999994476235309e-06,
+      "loss": 6.1114,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999987571532302e-06,
+      "loss": 5.5645,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999977904953434e-06,
+      "loss": 5.2072,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999965476504043e-06,
+      "loss": 4.9137,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999950286190993e-06,
+      "loss": 4.702,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999932334022677e-06,
+      "loss": 4.4843,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.99991162000901e-06,
+      "loss": 4.3326,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999888144161436e-06,
+      "loss": 4.158,
+      "step": 45
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999861906492918e-06,
+      "loss": 3.9941,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999832907017953e-06,
+      "loss": 3.8764,
+      "step": 55
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999801145752559e-06,
+      "loss": 3.8076,
+      "step": 60
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999766622714278e-06,
+      "loss": 3.6817,
+      "step": 65
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999729337922183e-06,
+      "loss": 3.6254,
+      "step": 70
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999689291396868e-06,
+      "loss": 3.5715,
+      "step": 75
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.99964648316045e-06,
+      "loss": 3.4705,
+      "step": 80
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999600913236582e-06,
+      "loss": 3.4141,
+      "step": 85
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999552581650432e-06,
+      "loss": 3.3455,
+      "step": 90
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999501488428695e-06,
+      "loss": 3.2984,
+      "step": 95
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999447633599597e-06,
+      "loss": 3.2459,
+      "step": 100
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999391017192888e-06,
+      "loss": 3.2819,
+      "step": 105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999331639239838e-06,
+      "loss": 3.1869,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999269499773245e-06,
+      "loss": 3.1485,
+      "step": 115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999204598827437e-06,
+      "loss": 3.1166,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999136936438261e-06,
+      "loss": 3.0867,
+      "step": 125
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999066512643094e-06,
+      "loss": 3.0882,
+      "step": 130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998993327480836e-06,
+      "loss": 3.0295,
+      "step": 135
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998917380991913e-06,
+      "loss": 3.0377,
+      "step": 140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998838673218273e-06,
+      "loss": 2.9933,
+      "step": 145
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998757204203398e-06,
+      "loss": 2.9961,
+      "step": 150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998672973992284e-06,
+      "loss": 2.962,
+      "step": 155
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998585982631464e-06,
+      "loss": 2.974,
+      "step": 160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998496230168982e-06,
+      "loss": 2.9417,
+      "step": 165
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998403716654421e-06,
+      "loss": 2.9106,
+      "step": 170
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998308442138882e-06,
+      "loss": 2.9218,
+      "step": 175
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998210406674991e-06,
+      "loss": 2.8947,
+      "step": 180
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998109610316903e-06,
+      "loss": 2.8701,
+      "step": 185
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998006053120292e-06,
+      "loss": 2.8695,
+      "step": 190
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997899735142365e-06,
+      "loss": 2.8969,
+      "step": 195
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997790656441846e-06,
+      "loss": 2.8447,
+      "step": 200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99767881707899e-06,
+      "loss": 2.8303,
+      "step": 205
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997564217115572e-06,
+      "loss": 2.8372,
+      "step": 210
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997446856614895e-06,
+      "loss": 2.8378,
+      "step": 215
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997326735641788e-06,
+      "loss": 2.8146,
+      "step": 220
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.9972038542626e-06,
+      "loss": 2.8088,
+      "step": 225
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997078212545211e-06,
+      "loss": 2.7869,
+      "step": 230
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99694981055902e-06,
+      "loss": 2.743,
+      "step": 235
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996818648374956e-06,
+      "loss": 2.7894,
+      "step": 240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996684726065467e-06,
+      "loss": 2.7621,
+      "step": 245
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99654804370453e-06,
+      "loss": 2.745,
+      "step": 250
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996408601367644e-06,
+      "loss": 2.7576,
+      "step": 255
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996266399131835e-06,
+      "loss": 2.7479,
+      "step": 260
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996121437075652e-06,
+      "loss": 2.7154,
+      "step": 265
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99597371527917e-06,
+      "loss": 2.7631,
+      "step": 270
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995823233823982e-06,
+      "loss": 2.7356,
+      "step": 275
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995669992793216e-06,
+      "loss": 2.7183,
+      "step": 280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995513992271517e-06,
+      "loss": 2.729,
+      "step": 285
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995355232345054e-06,
+      "loss": 2.7036,
+      "step": 290
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995193713101525e-06,
+      "loss": 2.681,
+      "step": 295
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995029434630148e-06,
+      "loss": 2.6817,
+      "step": 300
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.994862397021667e-06,
+      "loss": 2.6865,
+      "step": 305
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99469260036835e-06,
+      "loss": 2.6634,
+      "step": 310
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.994520044763989e-06,
+      "loss": 2.6508,
+      "step": 315
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994344730303898e-06,
+      "loss": 2.6425,
+      "step": 320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994166657084916e-06,
+      "loss": 2.6484,
+      "step": 325
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99398582520541e-06,
+      "loss": 2.6713,
+      "step": 330
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993802234765264e-06,
+      "loss": 2.6081,
+      "step": 335
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99361588586589e-06,
+      "loss": 2.6069,
+      "step": 340
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993426778610225e-06,
+      "loss": 2.6541,
+      "step": 345
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993234913102726e-06,
+      "loss": 2.6448,
+      "step": 350
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993040289449371e-06,
+      "loss": 2.6292,
+      "step": 355
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992842907757671e-06,
+      "loss": 2.6578,
+      "step": 360
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992642768136653e-06,
+      "loss": 2.628,
+      "step": 365
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99243987069687e-06,
+      "loss": 2.5677,
+      "step": 370
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992234215550399e-06,
+      "loss": 2.5979,
+      "step": 375
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992025802810835e-06,
+      "loss": 2.5936,
+      "step": 380
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991814632593302e-06,
+      "loss": 2.6004,
+      "step": 385
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991600705014448e-06,
+      "loss": 2.5807,
+      "step": 390
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991384020192439e-06,
+      "loss": 2.5749,
+      "step": 395
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991164578246965e-06,
+      "loss": 2.5875,
+      "step": 400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990942379299246e-06,
+      "loss": 2.5778,
+      "step": 405
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990717423472017e-06,
+      "loss": 2.5949,
+      "step": 410
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990489710889535e-06,
+      "loss": 2.5735,
+      "step": 415
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990259241677586e-06,
+      "loss": 2.5611,
+      "step": 420
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990026015963478e-06,
+      "loss": 2.5709,
+      "step": 425
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989790033876036e-06,
+      "loss": 2.5608,
+      "step": 430
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989551295545612e-06,
+      "loss": 2.5685,
+      "step": 435
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989309801104079e-06,
+      "loss": 2.5568,
+      "step": 440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989065550684833e-06,
+      "loss": 2.5278,
+      "step": 445
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988818544422794e-06,
+      "loss": 2.545,
+      "step": 450
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.9885687824544e-06,
+      "loss": 2.5469,
+      "step": 455
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988316264917614e-06,
+      "loss": 2.5342,
+      "step": 460
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988060991951922e-06,
+      "loss": 2.5203,
+      "step": 465
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.98780296369833e-06,
+      "loss": 2.5267,
+      "step": 470
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987542180299367e-06,
+      "loss": 2.5116,
+      "step": 475
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987278641899086e-06,
+      "loss": 2.5176,
+      "step": 480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987012348643054e-06,
+      "loss": 2.5295,
+      "step": 485
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.986743300678368e-06,
+      "loss": 2.5254,
+      "step": 490
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.986471498153646e-06,
+      "loss": 2.5357,
+      "step": 495
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.986196941219022e-06,
+      "loss": 2.5019,
+      "step": 500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.985919630026158e-06,
+      "loss": 2.554,
+      "step": 505
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.98563956472823e-06,
+      "loss": 2.4973,
+      "step": 510
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.985356745479942e-06,
+      "loss": 2.5188,
+      "step": 515
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.985071172437517e-06,
+      "loss": 2.5064,
+      "step": 520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.984782845758698e-06,
+      "loss": 2.4875,
+      "step": 525
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.98449176560275e-06,
+      "loss": 2.5104,
+      "step": 530
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984197932130461e-06,
+      "loss": 2.4804,
+      "step": 535
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983901345504135e-06,
+      "loss": 2.4735,
+      "step": 540
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.9836020058876e-06,
+      "loss": 2.4878,
+      "step": 545
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983299913446206e-06,
+      "loss": 2.4698,
+      "step": 550
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.982995068346816e-06,
+      "loss": 2.4937,
+      "step": 555
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.982687470757828e-06,
+      "loss": 2.4839,
+      "step": 560
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.982377120849145e-06,
+      "loss": 2.4931,
+      "step": 565
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.9820640187922e-06,
+      "loss": 2.4804,
+      "step": 570
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.981748164759944e-06,
+      "loss": 2.5001,
+      "step": 575
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.981429558926843e-06,
+      "loss": 2.4816,
+      "step": 580
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.981108201468892e-06,
+      "loss": 2.4846,
+      "step": 585
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.9807840925636e-06,
+      "loss": 2.4692,
+      "step": 590
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.980457232389995e-06,
+      "loss": 2.4913,
+      "step": 595
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.98012762112863e-06,
+      "loss": 2.4675,
+      "step": 600
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.979795258961574e-06,
+      "loss": 2.455,
+      "step": 605
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.979460146072413e-06,
+      "loss": 2.4365,
+      "step": 610
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.979122282646259e-06,
+      "loss": 2.4343,
+      "step": 615
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.97878166886974e-06,
+      "loss": 2.4389,
+      "step": 620
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.978438304931e-06,
+      "loss": 2.4318,
+      "step": 625
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.978092191019706e-06,
+      "loss": 2.42,
+      "step": 630
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.977743327327045e-06,
+      "loss": 2.448,
+      "step": 635
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.97739171404572e-06,
+      "loss": 2.4654,
+      "step": 640
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.977037351369954e-06,
+      "loss": 2.4282,
+      "step": 645
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.976680239495489e-06,
+      "loss": 2.4549,
+      "step": 650
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.976320378619583e-06,
+      "loss": 2.4481,
+      "step": 655
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.97595776894102e-06,
+      "loss": 2.4155,
+      "step": 660
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.975592410660092e-06,
+      "loss": 2.4541,
+      "step": 665
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.975224303978617e-06,
+      "loss": 2.4337,
+      "step": 670
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.974853449099926e-06,
+      "loss": 2.4335,
+      "step": 675
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.974479846228872e-06,
+      "loss": 2.4388,
+      "step": 680
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.974103495571826e-06,
+      "loss": 2.4307,
+      "step": 685
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.973724397336674e-06,
+      "loss": 2.4419,
+      "step": 690
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.97334255173282e-06,
+      "loss": 2.4211,
+      "step": 695
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.972957958971187e-06,
+      "loss": 2.3968,
+      "step": 700
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.972570619264215e-06,
+      "loss": 2.3951,
+      "step": 705
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.972180532825863e-06,
+      "loss": 2.4216,
+      "step": 710
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.971787699871604e-06,
+      "loss": 2.4134,
+      "step": 715
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.97139212061843e-06,
+      "loss": 2.4145,
+      "step": 720
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.97099379528485e-06,
+      "loss": 2.4184,
+      "step": 725
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.97059272409089e-06,
+      "loss": 2.4229,
+      "step": 730
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.97018890725809e-06,
+      "loss": 2.4093,
+      "step": 735
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.969782345009513e-06,
+      "loss": 2.4095,
+      "step": 740
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.969373037569729e-06,
+      "loss": 2.4073,
+      "step": 745
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.968960985164835e-06,
+      "loss": 2.4169,
+      "step": 750
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.968546188022436e-06,
+      "loss": 2.3755,
+      "step": 755
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.968128646371657e-06,
+      "loss": 2.397,
+      "step": 760
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.96770836044314e-06,
+      "loss": 2.3995,
+      "step": 765
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.967285330469038e-06,
+      "loss": 2.3782,
+      "step": 770
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.966859556683025e-06,
+      "loss": 2.3683,
+      "step": 775
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.966431039320287e-06,
+      "loss": 2.3804,
+      "step": 780
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.965999778617528e-06,
+      "loss": 2.3983,
+      "step": 785
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.965565774812966e-06,
+      "loss": 2.3802,
+      "step": 790
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.965129028146337e-06,
+      "loss": 2.3667,
+      "step": 795
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.964689538858884e-06,
+      "loss": 2.4291,
+      "step": 800
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.964247307193375e-06,
+      "loss": 2.3868,
+      "step": 805
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.963802333394086e-06,
+      "loss": 2.365,
+      "step": 810
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.963354617706813e-06,
+      "loss": 2.3828,
+      "step": 815
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.96290416037886e-06,
+      "loss": 2.3892,
+      "step": 820
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.962450961659051e-06,
+      "loss": 2.386,
+      "step": 825
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.961995021797724e-06,
+      "loss": 2.374,
+      "step": 830
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.961536341046726e-06,
+      "loss": 2.3785,
+      "step": 835
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.961074919659427e-06,
+      "loss": 2.3724,
+      "step": 840
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.960610757890697e-06,
+      "loss": 2.3675,
+      "step": 845
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.960143855996935e-06,
+      "loss": 2.3889,
+      "step": 850
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.959674214236043e-06,
+      "loss": 2.3601,
+      "step": 855
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.959201832867442e-06,
+      "loss": 2.3484,
+      "step": 860
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.958726712152064e-06,
+      "loss": 2.3407,
+      "step": 865
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.958248852352355e-06,
+      "loss": 2.3561,
+      "step": 870
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.957768253732272e-06,
+      "loss": 2.3704,
+      "step": 875
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.957284916557288e-06,
+      "loss": 2.3471,
+      "step": 880
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.956798841094384e-06,
+      "loss": 2.3548,
+      "step": 885
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.956310027612062e-06,
+      "loss": 2.3365,
+      "step": 890
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.955818476380326e-06,
+      "loss": 2.3598,
+      "step": 895
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.955324187670702e-06,
+      "loss": 2.3522,
+      "step": 900
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.95482716175622e-06,
+      "loss": 2.3431,
+      "step": 905
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.954327398911425e-06,
+      "loss": 2.3278,
+      "step": 910
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.953824899412379e-06,
+      "loss": 2.3523,
+      "step": 915
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.953319663536644e-06,
+      "loss": 2.3411,
+      "step": 920
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.952811691563307e-06,
+      "loss": 2.3565,
+      "step": 925
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.952300983772955e-06,
+      "loss": 2.3311,
+      "step": 930
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.951787540447694e-06,
+      "loss": 2.3516,
+      "step": 935
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.951271361871138e-06,
+      "loss": 2.3449,
+      "step": 940
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.95075244832841e-06,
+      "loss": 2.3244,
+      "step": 945
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.950230800106147e-06,
+      "loss": 2.3651,
+      "step": 950
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.949706417492494e-06,
+      "loss": 2.3038,
+      "step": 955
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.94917930077711e-06,
+      "loss": 2.3587,
+      "step": 960
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.94864945025116e-06,
+      "loss": 2.354,
+      "step": 965
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.94811686620732e-06,
+      "loss": 2.3128,
+      "step": 970
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.94758154893978e-06,
+      "loss": 2.3102,
+      "step": 975
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.947043498744235e-06,
+      "loss": 2.3381,
+      "step": 980
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.94650271591789e-06,
+      "loss": 2.3278,
+      "step": 985
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.945959200759464e-06,
+      "loss": 2.3503,
+      "step": 990
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.94541295356918e-06,
+      "loss": 2.336,
+      "step": 995
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.944863974648772e-06,
+      "loss": 2.3124,
+      "step": 1000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.944312264301483e-06,
+      "loss": 2.3076,
+      "step": 1005
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.943757822832066e-06,
+      "loss": 2.3211,
+      "step": 1010
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.943200650546779e-06,
+      "loss": 2.32,
+      "step": 1015
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.942640747753394e-06,
+      "loss": 2.3418,
+      "step": 1020
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.942078114761186e-06,
+      "loss": 2.3564,
+      "step": 1025
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.941512751880941e-06,
+      "loss": 2.3392,
+      "step": 1030
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.940944659424953e-06,
+      "loss": 2.3173,
+      "step": 1035
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.940373837707023e-06,
+      "loss": 2.3268,
+      "step": 1040
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.939800287042457e-06,
+      "loss": 2.2995,
+      "step": 1045
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.939224007748073e-06,
+      "loss": 2.3241,
+      "step": 1050
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.938645000142195e-06,
+      "loss": 2.36,
+      "step": 1055
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.93806326454465e-06,
+      "loss": 2.3344,
+      "step": 1060
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.937478801276779e-06,
+      "loss": 2.3314,
+      "step": 1065
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.936891610661424e-06,
+      "loss": 2.3148,
+      "step": 1070
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.936301693022934e-06,
+      "loss": 2.3089,
+      "step": 1075
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.935709048687168e-06,
+      "loss": 2.3079,
+      "step": 1080
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.935113677981488e-06,
+      "loss": 2.2986,
+      "step": 1085
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.934515581234761e-06,
+      "loss": 2.3136,
+      "step": 1090
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.933914758777366e-06,
+      "loss": 2.3211,
+      "step": 1095
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.933311210941177e-06,
+      "loss": 2.3279,
+      "step": 1100
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.932704938059585e-06,
+      "loss": 2.288,
+      "step": 1105
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.932095940467479e-06,
+      "loss": 2.3248,
+      "step": 1110
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.931484218501255e-06,
+      "loss": 2.3064,
+      "step": 1115
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.930869772498814e-06,
+      "loss": 2.3049,
+      "step": 1120
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.930252602799561e-06,
+      "loss": 2.3026,
+      "step": 1125
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.929632709744405e-06,
+      "loss": 2.308,
+      "step": 1130
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.929010093675766e-06,
+      "loss": 2.3176,
+      "step": 1135
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.928384754937555e-06,
+      "loss": 2.3236,
+      "step": 1140
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.927756693875199e-06,
+      "loss": 2.3088,
+      "step": 1145
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.927125910835623e-06,
+      "loss": 2.2927,
+      "step": 1150
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.926492406167255e-06,
+      "loss": 2.2998,
+      "step": 1155
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.925856180220033e-06,
+      "loss": 2.3028,
+      "step": 1160
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.925217233345389e-06,
+      "loss": 2.3125,
+      "step": 1165
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.924575565896263e-06,
+      "loss": 2.3274,
+      "step": 1170
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.923931178227097e-06,
+      "loss": 2.296,
+      "step": 1175
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.923284070693839e-06,
+      "loss": 2.3133,
+      "step": 1180
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.92263424365393e-06,
+      "loss": 2.3098,
+      "step": 1185
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.921981697466325e-06,
+      "loss": 2.291,
+      "step": 1190
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.921326432491471e-06,
+      "loss": 2.2926,
+      "step": 1195
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.920668449091321e-06,
+      "loss": 2.2831,
+      "step": 1200
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.920007747629334e-06,
+      "loss": 2.2708,
+      "step": 1205
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.919344328470463e-06,
+      "loss": 2.2982,
+      "step": 1210
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.918678191981165e-06,
+      "loss": 2.3061,
+      "step": 1215
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.918009338529398e-06,
+      "loss": 2.2982,
+      "step": 1220
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.917337768484621e-06,
+      "loss": 2.2838,
+      "step": 1225
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.916663482217796e-06,
+      "loss": 2.2969,
+      "step": 1230
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.915986480101381e-06,
+      "loss": 2.2856,
+      "step": 1235
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.915306762509333e-06,
+      "loss": 2.2856,
+      "step": 1240
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.914624329817117e-06,
+      "loss": 2.3029,
+      "step": 1245
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.913939182401693e-06,
+      "loss": 2.2719,
+      "step": 1250
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.913251320641515e-06,
+      "loss": 2.2958,
+      "step": 1255
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.912560744916547e-06,
+      "loss": 2.2606,
+      "step": 1260
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.911867455608243e-06,
+      "loss": 2.2953,
+      "step": 1265
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.911171453099563e-06,
+      "loss": 2.2636,
+      "step": 1270
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.910472737774963e-06,
+      "loss": 2.274,
+      "step": 1275
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.909771310020393e-06,
+      "loss": 2.2968,
+      "step": 1280
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.909067170223307e-06,
+      "loss": 2.2779,
+      "step": 1285
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.908360318772656e-06,
+      "loss": 2.2929,
+      "step": 1290
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.907650756058888e-06,
+      "loss": 2.2809,
+      "step": 1295
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.906938482473947e-06,
+      "loss": 2.3105,
+      "step": 1300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.906223498411279e-06,
+      "loss": 2.275,
+      "step": 1305
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.905505804265822e-06,
+      "loss": 2.291,
+      "step": 1310
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.904785400434017e-06,
+      "loss": 2.2739,
+      "step": 1315
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.904062287313794e-06,
+      "loss": 2.2608,
+      "step": 1320
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.903336465304587e-06,
+      "loss": 2.2567,
+      "step": 1325
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.90260793480732e-06,
+      "loss": 2.2787,
+      "step": 1330
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.901876696224418e-06,
+      "loss": 2.2715,
+      "step": 1335
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.901142749959801e-06,
+      "loss": 2.2691,
+      "step": 1340
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.900406096418881e-06,
+      "loss": 2.2732,
+      "step": 1345
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.89966673600857e-06,
+      "loss": 2.3022,
+      "step": 1350
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.898924669137273e-06,
+      "loss": 2.2482,
+      "step": 1355
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.898179896214891e-06,
+      "loss": 2.2358,
+      "step": 1360
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.897432417652817e-06,
+      "loss": 2.2628,
+      "step": 1365
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.896682233863942e-06,
+      "loss": 2.233,
+      "step": 1370
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.895929345262651e-06,
+      "loss": 2.2797,
+      "step": 1375
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.89517375226482e-06,
+      "loss": 2.2743,
+      "step": 1380
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.894415455287823e-06,
+      "loss": 2.2731,
+      "step": 1385
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.893654454750523e-06,
+      "loss": 2.2853,
+      "step": 1390
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.89289075107328e-06,
+      "loss": 2.3017,
+      "step": 1395
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.892124344677947e-06,
+      "loss": 2.2678,
+      "step": 1400
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.89135523598787e-06,
+      "loss": 2.2714,
+      "step": 1405
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.890583425427881e-06,
+      "loss": 2.2688,
+      "step": 1410
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.889808913424315e-06,
+      "loss": 2.2587,
+      "step": 1415
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.889031700404994e-06,
+      "loss": 2.2442,
+      "step": 1420
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.88825178679923e-06,
+      "loss": 2.2516,
+      "step": 1425
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.88746917303783e-06,
+      "loss": 2.2498,
+      "step": 1430
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.886683859553092e-06,
+      "loss": 2.2178,
+      "step": 1435
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.885895846778804e-06,
+      "loss": 2.2741,
+      "step": 1440
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.885105135150248e-06,
+      "loss": 2.2779,
+      "step": 1445
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.88431172510419e-06,
+      "loss": 2.2663,
+      "step": 1450
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.883515617078897e-06,
+      "loss": 2.2562,
+      "step": 1455
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.882716811514115e-06,
+      "loss": 2.2267,
+      "step": 1460
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.881915308851086e-06,
+      "loss": 2.2354,
+      "step": 1465
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.881111109532547e-06,
+      "loss": 2.2377,
+      "step": 1470
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.880304214002712e-06,
+      "loss": 2.2625,
+      "step": 1475
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.879494622707296e-06,
+      "loss": 2.2636,
+      "step": 1480
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.878682336093495e-06,
+      "loss": 2.2395,
+      "step": 1485
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.877867354609998e-06,
+      "loss": 2.2749,
+      "step": 1490
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.877049678706982e-06,
+      "loss": 2.2574,
+      "step": 1495
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.876229308836112e-06,
+      "loss": 2.2486,
+      "step": 1500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.875406245450541e-06,
+      "loss": 2.2555,
+      "step": 1505
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.87458048900491e-06,
+      "loss": 2.2383,
+      "step": 1510
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.873752039955348e-06,
+      "loss": 2.2232,
+      "step": 1515
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.872920898759469e-06,
+      "loss": 2.2484,
+      "step": 1520
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.872087065876379e-06,
+      "loss": 2.2463,
+      "step": 1525
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.871250541766664e-06,
+      "loss": 2.2362,
+      "step": 1530
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.870411326892403e-06,
+      "loss": 2.2644,
+      "step": 1535
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.869569421717159e-06,
+      "loss": 2.2506,
+      "step": 1540
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.868724826705978e-06,
+      "loss": 2.271,
+      "step": 1545
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.867877542325395e-06,
+      "loss": 2.2528,
+      "step": 1550
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.867027569043432e-06,
+      "loss": 2.2448,
+      "step": 1555
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.866174907329593e-06,
+      "loss": 2.2388,
+      "step": 1560
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.865319557654868e-06,
+      "loss": 2.2709,
+      "step": 1565
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.864461520491734e-06,
+      "loss": 2.2296,
+      "step": 1570
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.86360079631415e-06,
+      "loss": 2.251,
+      "step": 1575
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.862737385597559e-06,
+      "loss": 2.2299,
+      "step": 1580
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.861871288818887e-06,
+      "loss": 2.2379,
+      "step": 1585
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.861002506456547e-06,
+      "loss": 2.2619,
+      "step": 1590
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.860131038990436e-06,
+      "loss": 2.2358,
+      "step": 1595
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.859256886901931e-06,
+      "loss": 2.2291,
+      "step": 1600
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.858380050673892e-06,
+      "loss": 2.2375,
+      "step": 1605
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.857500530790664e-06,
+      "loss": 2.1867,
+      "step": 1610
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.856618327738075e-06,
+      "loss": 2.2541,
+      "step": 1615
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.855733442003428e-06,
+      "loss": 2.2424,
+      "step": 1620
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.854845874075517e-06,
+      "loss": 2.2541,
+      "step": 1625
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.853955624444613e-06,
+      "loss": 2.229,
+      "step": 1630
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.853062693602468e-06,
+      "loss": 2.2463,
+      "step": 1635
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.852167082042317e-06,
+      "loss": 2.2421,
+      "step": 1640
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.851268790258876e-06,
+      "loss": 2.208,
+      "step": 1645
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.850367818748338e-06,
+      "loss": 2.228,
+      "step": 1650
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.84946416800838e-06,
+      "loss": 2.2298,
+      "step": 1655
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.848557838538157e-06,
+      "loss": 2.2317,
+      "step": 1660
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.847648830838304e-06,
+      "loss": 2.2104,
+      "step": 1665
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.846737145410935e-06,
+      "loss": 2.2293,
+      "step": 1670
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.845822782759645e-06,
+      "loss": 2.2394,
+      "step": 1675
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.844905743389506e-06,
+      "loss": 2.2296,
+      "step": 1680
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.84398602780707e-06,
+      "loss": 2.2333,
+      "step": 1685
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.843063636520362e-06,
+      "loss": 2.2132,
+      "step": 1690
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.842138570038895e-06,
+      "loss": 2.2542,
+      "step": 1695
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.841210828873652e-06,
+      "loss": 2.2266,
+      "step": 1700
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.840280413537094e-06,
+      "loss": 2.2104,
+      "step": 1705
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.839347324543162e-06,
+      "loss": 2.2283,
+      "step": 1710
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.83841156240727e-06,
+      "loss": 2.2305,
+      "step": 1715
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.837473127646315e-06,
+      "loss": 2.2238,
+      "step": 1720
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.836532020778666e-06,
+      "loss": 2.2217,
+      "step": 1725
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.835588242324166e-06,
+      "loss": 2.2232,
+      "step": 1730
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.834641792804136e-06,
+      "loss": 2.2005,
+      "step": 1735
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.833692672741373e-06,
+      "loss": 2.2158,
+      "step": 1740
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.83274088266015e-06,
+      "loss": 2.2305,
+      "step": 1745
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.831786423086211e-06,
+      "loss": 2.2378,
+      "step": 1750
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.830829294546782e-06,
+      "loss": 2.1993,
+      "step": 1755
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.829869497570552e-06,
+      "loss": 2.2027,
+      "step": 1760
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.828907032687695e-06,
+      "loss": 2.217,
+      "step": 1765
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.827941900429852e-06,
+      "loss": 2.2387,
+      "step": 1770
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.82697410133014e-06,
+      "loss": 2.2128,
+      "step": 1775
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.826003635923148e-06,
+      "loss": 2.2353,
+      "step": 1780
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.82503050474494e-06,
+      "loss": 2.2243,
+      "step": 1785
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.82405470833305e-06,
+      "loss": 2.2276,
+      "step": 1790
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.823076247226481e-06,
+      "loss": 2.2254,
+      "step": 1795
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.822095121965717e-06,
+      "loss": 2.2137,
+      "step": 1800
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.821111333092708e-06,
+      "loss": 2.198,
+      "step": 1805
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.820124881150875e-06,
+      "loss": 2.1839,
+      "step": 1810
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.81913576668511e-06,
+      "loss": 2.2021,
+      "step": 1815
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.81814399024178e-06,
+      "loss": 2.2349,
+      "step": 1820
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.817149552368715e-06,
+      "loss": 2.2219,
+      "step": 1825
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.816152453615222e-06,
+      "loss": 2.2091,
+      "step": 1830
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.81515269453207e-06,
+      "loss": 2.2293,
+      "step": 1835
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.814150275671511e-06,
+      "loss": 2.2136,
+      "step": 1840
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.813145197587251e-06,
+      "loss": 2.201,
+      "step": 1845
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.812137460834472e-06,
+      "loss": 2.2107,
+      "step": 1850
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.811127065969828e-06,
+      "loss": 2.2064,
+      "step": 1855
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.810114013551432e-06,
+      "loss": 2.2379,
+      "step": 1860
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.809098304138876e-06,
+      "loss": 2.2208,
+      "step": 1865
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.808079938293211e-06,
+      "loss": 2.2349,
+      "step": 1870
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.80705891657696e-06,
+      "loss": 2.2228,
+      "step": 1875
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.806035239554109e-06,
+      "loss": 2.2063,
+      "step": 1880
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.805008907790114e-06,
+      "loss": 2.1925,
+      "step": 1885
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.803979921851898e-06,
+      "loss": 2.2243,
+      "step": 1890
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.802948282307847e-06,
+      "loss": 2.1843,
+      "step": 1895
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.801913989727816e-06,
+      "loss": 2.1958,
+      "step": 1900
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.800877044683124e-06,
+      "loss": 2.1939,
+      "step": 1905
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.799837447746553e-06,
+      "loss": 2.2237,
+      "step": 1910
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.798795199492352e-06,
+      "loss": 2.1976,
+      "step": 1915
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.797750300496239e-06,
+      "loss": 2.2079,
+      "step": 1920
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.796702751335388e-06,
+      "loss": 2.1926,
+      "step": 1925
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.795652552588439e-06,
+      "loss": 2.1858,
+      "step": 1930
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.7945997048355e-06,
+      "loss": 2.2021,
+      "step": 1935
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.79354420865814e-06,
+      "loss": 2.183,
+      "step": 1940
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.792486064639388e-06,
+      "loss": 2.1943,
+      "step": 1945
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.79142527336374e-06,
+      "loss": 2.2179,
+      "step": 1950
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.790361835417151e-06,
+      "loss": 2.2202,
+      "step": 1955
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.789295751387038e-06,
+      "loss": 2.2173,
+      "step": 1960
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.788227021862282e-06,
+      "loss": 2.1831,
+      "step": 1965
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.787155647433227e-06,
+      "loss": 2.1832,
+      "step": 1970
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.78608162869167e-06,
+      "loss": 2.2089,
+      "step": 1975
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.785004966230876e-06,
+      "loss": 2.2493,
+      "step": 1980
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.78392566064557e-06,
+      "loss": 2.2085,
+      "step": 1985
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.782843712531932e-06,
+      "loss": 2.188,
+      "step": 1990
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.781759122487606e-06,
+      "loss": 2.2299,
+      "step": 1995
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.780671891111695e-06,
+      "loss": 2.1931,
+      "step": 2000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.779582019004759e-06,
+      "loss": 2.2105,
+      "step": 2005
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.778489506768817e-06,
+      "loss": 2.2072,
+      "step": 2010
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.77739435500735e-06,
+      "loss": 2.186,
+      "step": 2015
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.776296564325292e-06,
+      "loss": 2.2117,
+      "step": 2020
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.775196135329036e-06,
+      "loss": 2.1903,
+      "step": 2025
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.774093068626436e-06,
+      "loss": 2.1706,
+      "step": 2030
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.772987364826796e-06,
+      "loss": 2.1905,
+      "step": 2035
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.771879024540887e-06,
+      "loss": 2.1988,
+      "step": 2040
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.770768048380925e-06,
+      "loss": 2.1929,
+      "step": 2045
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.769654436960587e-06,
+      "loss": 2.1843,
+      "step": 2050
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.76853819089501e-06,
+      "loss": 2.1841,
+      "step": 2055
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.767419310800779e-06,
+      "loss": 2.1837,
+      "step": 2060
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.766297797295937e-06,
+      "loss": 2.1844,
+      "step": 2065
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.765173650999984e-06,
+      "loss": 2.1905,
+      "step": 2070
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.76404687253387e-06,
+      "loss": 2.1725,
+      "step": 2075
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.76291746252e-06,
+      "loss": 2.2227,
+      "step": 2080
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.761785421582238e-06,
+      "loss": 2.1653,
+      "step": 2085
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.760650750345893e-06,
+      "loss": 2.1827,
+      "step": 2090
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.759513449437733e-06,
+      "loss": 2.1915,
+      "step": 2095
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.758373519485973e-06,
+      "loss": 2.1821,
+      "step": 2100
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.757230961120288e-06,
+      "loss": 2.192,
+      "step": 2105
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.756085774971796e-06,
+      "loss": 2.1871,
+      "step": 2110
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.754937961673075e-06,
+      "loss": 2.1872,
+      "step": 2115
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.753787521858146e-06,
+      "loss": 2.2191,
+      "step": 2120
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.75263445616249e-06,
+      "loss": 2.1678,
+      "step": 2125
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.75147876522303e-06,
+      "loss": 2.1845,
+      "step": 2130
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.750320449678143e-06,
+      "loss": 2.197,
+      "step": 2135
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.749159510167655e-06,
+      "loss": 2.214,
+      "step": 2140
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.747995947332844e-06,
+      "loss": 2.1796,
+      "step": 2145
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.746829761816432e-06,
+      "loss": 2.1686,
+      "step": 2150
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.745660954262595e-06,
+      "loss": 2.1858,
+      "step": 2155
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.744489525316953e-06,
+      "loss": 2.1693,
+      "step": 2160
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.743315475626577e-06,
+      "loss": 2.1963,
+      "step": 2165
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.742138805839983e-06,
+      "loss": 2.2058,
+      "step": 2170
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.740959516607138e-06,
+      "loss": 2.1977,
+      "step": 2175
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.739777608579452e-06,
+      "loss": 2.2125,
+      "step": 2180
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.738593082409785e-06,
+      "loss": 2.1706,
+      "step": 2185
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.737405938752437e-06,
+      "loss": 2.1992,
+      "step": 2190
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.736216178263165e-06,
+      "loss": 2.1648,
+      "step": 2195
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.735023801599159e-06,
+      "loss": 2.1713,
+      "step": 2200
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.733828809419063e-06,
+      "loss": 2.1702,
+      "step": 2205
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.732631202382963e-06,
+      "loss": 2.1605,
+      "step": 2210
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.731430981152388e-06,
+      "loss": 2.1777,
+      "step": 2215
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.730228146390313e-06,
+      "loss": 2.1658,
+      "step": 2220
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.729022698761153e-06,
+      "loss": 2.1471,
+      "step": 2225
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.72781463893077e-06,
+      "loss": 2.182,
+      "step": 2230
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.72660396756647e-06,
+      "loss": 2.1884,
+      "step": 2235
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.725390685336998e-06,
+      "loss": 2.164,
+      "step": 2240
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.724174792912543e-06,
+      "loss": 2.1819,
+      "step": 2245
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.722956290964734e-06,
+      "loss": 2.1403,
+      "step": 2250
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.721735180166645e-06,
+      "loss": 2.2035,
+      "step": 2255
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.720511461192788e-06,
+      "loss": 2.1791,
+      "step": 2260
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.719285134719117e-06,
+      "loss": 2.1785,
+      "step": 2265
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.718056201423026e-06,
+      "loss": 2.1962,
+      "step": 2270
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.716824661983349e-06,
+      "loss": 2.2125,
+      "step": 2275
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.715590517080358e-06,
+      "loss": 2.1394,
+      "step": 2280
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.714353767395768e-06,
+      "loss": 2.1933,
+      "step": 2285
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.713114413612729e-06,
+      "loss": 2.1315,
+      "step": 2290
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.711872456415832e-06,
+      "loss": 2.1698,
+      "step": 2295
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.710627896491103e-06,
+      "loss": 2.1386,
+      "step": 2300
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.709380734526009e-06,
+      "loss": 2.1544,
+      "step": 2305
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.708130971209454e-06,
+      "loss": 2.1756,
+      "step": 2310
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.706878607231777e-06,
+      "loss": 2.1812,
+      "step": 2315
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.705623643284754e-06,
+      "loss": 2.1844,
+      "step": 2320
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.704366080061598e-06,
+      "loss": 2.1969,
+      "step": 2325
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.70310591825696e-06,
+      "loss": 2.1654,
+      "step": 2330
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.70184315856692e-06,
+      "loss": 2.1566,
+      "step": 2335
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.700577801688997e-06,
+      "loss": 2.1522,
+      "step": 2340
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.699309848322147e-06,
+      "loss": 2.1648,
+      "step": 2345
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.698039299166758e-06,
+      "loss": 2.1866,
+      "step": 2350
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.696766154924646e-06,
+      "loss": 2.176,
+      "step": 2355
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.695490416299072e-06,
+      "loss": 2.1683,
+      "step": 2360
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.694212083994723e-06,
+      "loss": 2.1805,
+      "step": 2365
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.692931158717718e-06,
+      "loss": 2.1734,
+      "step": 2370
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.691647641175612e-06,
+      "loss": 2.1799,
+      "step": 2375
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.690361532077388e-06,
+      "loss": 2.1804,
+      "step": 2380
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.689072832133463e-06,
+      "loss": 2.1537,
+      "step": 2385
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.687781542055686e-06,
+      "loss": 2.1587,
+      "step": 2390
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.686487662557334e-06,
+      "loss": 2.1617,
+      "step": 2395
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.685191194353116e-06,
+      "loss": 2.1583,
+      "step": 2400
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.683892138159172e-06,
+      "loss": 2.1599,
+      "step": 2405
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.682590494693068e-06,
+      "loss": 2.1725,
+      "step": 2410
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.681286264673803e-06,
+      "loss": 2.1449,
+      "step": 2415
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.679979448821801e-06,
+      "loss": 2.1825,
+      "step": 2420
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.678670047858917e-06,
+      "loss": 2.1675,
+      "step": 2425
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.677358062508438e-06,
+      "loss": 2.1793,
+      "step": 2430
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.676043493495066e-06,
+      "loss": 2.1577,
+      "step": 2435
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.674726341544945e-06,
+      "loss": 2.1784,
+      "step": 2440
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.673406607385633e-06,
+      "loss": 2.1484,
+      "step": 2445
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.672084291746126e-06,
+      "loss": 2.1571,
+      "step": 2450
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.670759395356836e-06,
+      "loss": 2.2041,
+      "step": 2455
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.669431918949606e-06,
+      "loss": 2.1726,
+      "step": 2460
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.668101863257703e-06,
+      "loss": 2.1538,
+      "step": 2465
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.66676922901582e-06,
+      "loss": 2.1319,
+      "step": 2470
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.66543401696007e-06,
+      "loss": 2.1289,
+      "step": 2475
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.664096227827992e-06,
+      "loss": 2.1598,
+      "step": 2480
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.662755862358555e-06,
+      "loss": 2.1723,
+      "step": 2485
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.661412921292141e-06,
+      "loss": 2.1585,
+      "step": 2490
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.660067405370559e-06,
+      "loss": 2.1667,
+      "step": 2495
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.658719315337042e-06,
+      "loss": 2.143,
+      "step": 2500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.657368651936242e-06,
+      "loss": 2.1703,
+      "step": 2505
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.656015415914237e-06,
+      "loss": 2.1372,
+      "step": 2510
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.654659608018519e-06,
+      "loss": 2.1528,
+      "step": 2515
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.653301228998006e-06,
+      "loss": 2.1828,
+      "step": 2520
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.651940279603034e-06,
+      "loss": 2.2065,
+      "step": 2525
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.650576760585362e-06,
+      "loss": 2.1671,
+      "step": 2530
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.649210672698162e-06,
+      "loss": 2.1718,
+      "step": 2535
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.647842016696032e-06,
+      "loss": 2.1561,
+      "step": 2540
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.646470793334985e-06,
+      "loss": 2.1605,
+      "step": 2545
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.645097003372452e-06,
+      "loss": 2.1326,
+      "step": 2550
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.643720647567281e-06,
+      "loss": 2.1698,
+      "step": 2555
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.642341726679743e-06,
+      "loss": 2.1429,
+      "step": 2560
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.640960241471516e-06,
+      "loss": 2.1397,
+      "step": 2565
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.639576192705704e-06,
+      "loss": 2.1763,
+      "step": 2570
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.63818958114682e-06,
+      "loss": 2.1748,
+      "step": 2575
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.6368004075608e-06,
+      "loss": 2.1734,
+      "step": 2580
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.635408672714988e-06,
+      "loss": 2.1744,
+      "step": 2585
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.634014377378145e-06,
+      "loss": 2.1633,
+      "step": 2590
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.63261752232045e-06,
+      "loss": 2.1576,
+      "step": 2595
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.631218108313487e-06,
+      "loss": 2.1451,
+      "step": 2600
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.629816136130268e-06,
+      "loss": 2.1518,
+      "step": 2605
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.628411606545202e-06,
+      "loss": 2.136,
+      "step": 2610
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.627004520334122e-06,
+      "loss": 2.1625,
+      "step": 2615
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.62559487827427e-06,
+      "loss": 2.1851,
+      "step": 2620
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.624182681144298e-06,
+      "loss": 2.1372,
+      "step": 2625
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.622767929724271e-06,
+      "loss": 2.1446,
+      "step": 2630
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.621350624795662e-06,
+      "loss": 2.1605,
+      "step": 2635
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.61993076714136e-06,
+      "loss": 2.1717,
+      "step": 2640
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.61850835754566e-06,
+      "loss": 2.1592,
+      "step": 2645
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.617083396794268e-06,
+      "loss": 2.1662,
+      "step": 2650
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.615655885674298e-06,
+      "loss": 2.1436,
+      "step": 2655
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.614225824974275e-06,
+      "loss": 2.1602,
+      "step": 2660
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.612793215484131e-06,
+      "loss": 2.1214,
+      "step": 2665
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.611358057995203e-06,
+      "loss": 2.1471,
+      "step": 2670
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.609920353300243e-06,
+      "loss": 2.1452,
+      "step": 2675
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.608480102193403e-06,
+      "loss": 2.1222,
+      "step": 2680
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.607037305470241e-06,
+      "loss": 2.177,
+      "step": 2685
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.60559196392773e-06,
+      "loss": 2.1399,
+      "step": 2690
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.60414407836424e-06,
+      "loss": 2.1413,
+      "step": 2695
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.602693649579546e-06,
+      "loss": 2.1646,
+      "step": 2700
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.601240678374835e-06,
+      "loss": 2.1606,
+      "step": 2705
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.599785165552693e-06,
+      "loss": 2.146,
+      "step": 2710
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.59832711191711e-06,
+      "loss": 2.1346,
+      "step": 2715
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.596866518273482e-06,
+      "loss": 2.1421,
+      "step": 2720
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.595403385428605e-06,
+      "loss": 2.1256,
+      "step": 2725
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.593937714190682e-06,
+      "loss": 2.1328,
+      "step": 2730
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.592469505369315e-06,
+      "loss": 2.1519,
+      "step": 2735
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.590998759775506e-06,
+      "loss": 2.1681,
+      "step": 2740
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.58952547822166e-06,
+      "loss": 2.1665,
+      "step": 2745
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.588049661521586e-06,
+      "loss": 2.1585,
+      "step": 2750
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.586571310490489e-06,
+      "loss": 2.1464,
+      "step": 2755
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.585090425944975e-06,
+      "loss": 2.1489,
+      "step": 2760
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.58360700870305e-06,
+      "loss": 2.1337,
+      "step": 2765
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.58212105958412e-06,
+      "loss": 2.1376,
+      "step": 2770
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.580632579408988e-06,
+      "loss": 2.1505,
+      "step": 2775
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.579141568999854e-06,
+      "loss": 2.1579,
+      "step": 2780
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.57764802918032e-06,
+      "loss": 2.1446,
+      "step": 2785
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.57615196077538e-06,
+      "loss": 2.1273,
+      "step": 2790
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.574653364611429e-06,
+      "loss": 2.142,
+      "step": 2795
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.573152241516255e-06,
+      "loss": 2.1322,
+      "step": 2800
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.571648592319043e-06,
+      "loss": 2.1277,
+      "step": 2805
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.570142417850374e-06,
+      "loss": 2.1515,
+      "step": 2810
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.568633718942223e-06,
+      "loss": 2.1638,
+      "step": 2815
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.56712249642796e-06,
+      "loss": 2.1278,
+      "step": 2820
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.56560875114235e-06,
+      "loss": 2.1484,
+      "step": 2825
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.564092483921551e-06,
+      "loss": 2.1661,
+      "step": 2830
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.562573695603111e-06,
+      "loss": 2.1637,
+      "step": 2835
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.561052387025973e-06,
+      "loss": 2.1193,
+      "step": 2840
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.559528559030475e-06,
+      "loss": 2.141,
+      "step": 2845
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.558002212458341e-06,
+      "loss": 2.1504,
+      "step": 2850
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.556473348152691e-06,
+      "loss": 2.1186,
+      "step": 2855
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.554941966958033e-06,
+      "loss": 2.1345,
+      "step": 2860
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.553408069720267e-06,
+      "loss": 2.1315,
+      "step": 2865
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.551871657286679e-06,
+      "loss": 2.1828,
+      "step": 2870
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.550332730505949e-06,
+      "loss": 2.1249,
+      "step": 2875
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.548791290228146e-06,
+      "loss": 2.1379,
+      "step": 2880
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.547247337304723e-06,
+      "loss": 2.1302,
+      "step": 2885
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.545700872588523e-06,
+      "loss": 2.0922,
+      "step": 2890
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.54415189693378e-06,
+      "loss": 2.1352,
+      "step": 2895
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.542600411196107e-06,
+      "loss": 2.1248,
+      "step": 2900
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.541046416232512e-06,
+      "loss": 2.1366,
+      "step": 2905
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.539489912901384e-06,
+      "loss": 2.1288,
+      "step": 2910
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.5379309020625e-06,
+      "loss": 2.1316,
+      "step": 2915
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.53636938457702e-06,
+      "loss": 2.1295,
+      "step": 2920
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.53480536130749e-06,
+      "loss": 2.1314,
+      "step": 2925
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.533238833117839e-06,
+      "loss": 2.1583,
+      "step": 2930
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.531669800873383e-06,
+      "loss": 2.1264,
+      "step": 2935
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.530098265440815e-06,
+      "loss": 2.1308,
+      "step": 2940
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.528524227688215e-06,
+      "loss": 2.1506,
+      "step": 2945
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.526947688485048e-06,
+      "loss": 2.1389,
+      "step": 2950
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.525368648702156e-06,
+      "loss": 2.1065,
+      "step": 2955
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.523787109211759e-06,
+      "loss": 2.1165,
+      "step": 2960
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.522203070887467e-06,
+      "loss": 2.1218,
+      "step": 2965
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.520616534604266e-06,
+      "loss": 2.0804,
+      "step": 2970
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.519027501238518e-06,
+      "loss": 2.1611,
+      "step": 2975
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.51743597166797e-06,
+      "loss": 2.1311,
+      "step": 2980
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.515841946771746e-06,
+      "loss": 2.1461,
+      "step": 2985
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.514245427430346e-06,
+      "loss": 2.1391,
+      "step": 2990
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.512646414525651e-06,
+      "loss": 2.1535,
+      "step": 2995
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.511044908940918e-06,
+      "loss": 2.1323,
+      "step": 3000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.509440911560783e-06,
+      "loss": 2.1421,
+      "step": 3005
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.507834423271255e-06,
+      "loss": 2.1309,
+      "step": 3010
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.506225444959718e-06,
+      "loss": 2.1406,
+      "step": 3015
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.504613977514937e-06,
+      "loss": 2.1251,
+      "step": 3020
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.503000021827048e-06,
+      "loss": 2.1335,
+      "step": 3025
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.501383578787562e-06,
+      "loss": 2.1218,
+      "step": 3030
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.499764649289365e-06,
+      "loss": 2.1174,
+      "step": 3035
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.498143234226715e-06,
+      "loss": 2.1225,
+      "step": 3040
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.496519334495244e-06,
+      "loss": 2.141,
+      "step": 3045
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.494892950991956e-06,
+      "loss": 2.1257,
+      "step": 3050
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.493264084615228e-06,
+      "loss": 2.1237,
+      "step": 3055
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.491632736264804e-06,
+      "loss": 2.1149,
+      "step": 3060
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.489998906841808e-06,
+      "loss": 2.1577,
+      "step": 3065
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.488362597248725e-06,
+      "loss": 2.1435,
+      "step": 3070
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.486723808389415e-06,
+      "loss": 2.1121,
+      "step": 3075
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.485082541169107e-06,
+      "loss": 2.1218,
+      "step": 3080
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.4834387964944e-06,
+      "loss": 2.0987,
+      "step": 3085
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.481792575273256e-06,
+      "loss": 2.1351,
+      "step": 3090
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.480143878415012e-06,
+      "loss": 2.1349,
+      "step": 3095
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.478492706830369e-06,
+      "loss": 2.1062,
+      "step": 3100
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.476839061431395e-06,
+      "loss": 2.1035,
+      "step": 3105
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.475182943131526e-06,
+      "loss": 2.1316,
+      "step": 3110
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.473524352845561e-06,
+      "loss": 2.1144,
+      "step": 3115
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.471863291489669e-06,
+      "loss": 2.1185,
+      "step": 3120
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.470199759981378e-06,
+      "loss": 2.1035,
+      "step": 3125
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.468533759239586e-06,
+      "loss": 2.1503,
+      "step": 3130
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.466865290184553e-06,
+      "loss": 2.123,
+      "step": 3135
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.465194353737902e-06,
+      "loss": 2.148,
+      "step": 3140
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.463520950822618e-06,
+      "loss": 2.1182,
+      "step": 3145
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.46184508236305e-06,
+      "loss": 2.1691,
+      "step": 3150
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.46016674928491e-06,
+      "loss": 2.1171,
+      "step": 3155
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.458485952515266e-06,
+      "loss": 2.1481,
+      "step": 3160
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.456802692982555e-06,
+      "loss": 2.1112,
+      "step": 3165
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.455116971616566e-06,
+      "loss": 2.1159,
+      "step": 3170
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.453428789348455e-06,
+      "loss": 2.119,
+      "step": 3175
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.451738147110734e-06,
+      "loss": 2.1351,
+      "step": 3180
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.450045045837273e-06,
+      "loss": 2.1504,
+      "step": 3185
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.448349486463303e-06,
+      "loss": 2.1235,
+      "step": 3190
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.44665146992541e-06,
+      "loss": 2.1004,
+      "step": 3195
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.444950997161538e-06,
+      "loss": 2.1471,
+      "step": 3200
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.44324806911099e-06,
+      "loss": 2.1292,
+      "step": 3205
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.441542686714423e-06,
+      "loss": 2.1141,
+      "step": 3210
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.439834850913848e-06,
+      "loss": 2.1017,
+      "step": 3215
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.438124562652636e-06,
+      "loss": 2.112,
+      "step": 3220
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.436411822875508e-06,
+      "loss": 2.1026,
+      "step": 3225
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.434696632528544e-06,
+      "loss": 2.1321,
+      "step": 3230
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.432978992559172e-06,
+      "loss": 2.1044,
+      "step": 3235
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.431258903916179e-06,
+      "loss": 2.0905,
+      "step": 3240
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.4295363675497e-06,
+      "loss": 2.1084,
+      "step": 3245
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.427811384411224e-06,
+      "loss": 2.1071,
+      "step": 3250
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.426083955453591e-06,
+      "loss": 2.1144,
+      "step": 3255
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.424354081630993e-06,
+      "loss": 2.1213,
+      "step": 3260
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.422621763898968e-06,
+      "loss": 2.117,
+      "step": 3265
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.420887003214415e-06,
+      "loss": 2.1173,
+      "step": 3270
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.419149800535568e-06,
+      "loss": 2.1399,
+      "step": 3275
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.41741015682202e-06,
+      "loss": 2.1139,
+      "step": 3280
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.415668073034709e-06,
+      "loss": 2.0991,
+      "step": 3285
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.413923550135923e-06,
+      "loss": 2.1018,
+      "step": 3290
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.412176589089292e-06,
+      "loss": 2.1217,
+      "step": 3295
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.4104271908598e-06,
+      "loss": 2.1104,
+      "step": 3300
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.408675356413772e-06,
+      "loss": 2.1418,
+      "step": 3305
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.40692108671888e-06,
+      "loss": 2.1124,
+      "step": 3310
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.40516438274414e-06,
+      "loss": 2.0934,
+      "step": 3315
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.403405245459917e-06,
+      "loss": 2.0765,
+      "step": 3320
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.401643675837915e-06,
+      "loss": 2.1414,
+      "step": 3325
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.399879674851185e-06,
+      "loss": 2.1066,
+      "step": 3330
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.39811324347412e-06,
+      "loss": 2.1231,
+      "step": 3335
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.396344382682454e-06,
+      "loss": 2.0848,
+      "step": 3340
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.394573093453265e-06,
+      "loss": 2.1226,
+      "step": 3345
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.392799376764971e-06,
+      "loss": 2.1237,
+      "step": 3350
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.391023233597331e-06,
+      "loss": 2.12,
+      "step": 3355
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.389244664931447e-06,
+      "loss": 2.1196,
+      "step": 3360
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.387463671749757e-06,
+      "loss": 2.1182,
+      "step": 3365
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.385680255036041e-06,
+      "loss": 2.1253,
+      "step": 3370
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.383894415775414e-06,
+      "loss": 2.1624,
+      "step": 3375
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.382106154954335e-06,
+      "loss": 2.0965,
+      "step": 3380
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.380315473560595e-06,
+      "loss": 2.1111,
+      "step": 3385
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.378522372583326e-06,
+      "loss": 2.1251,
+      "step": 3390
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.376726853012994e-06,
+      "loss": 2.1048,
+      "step": 3395
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.374928915841402e-06,
+      "loss": 2.1157,
+      "step": 3400
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.373128562061687e-06,
+      "loss": 2.1144,
+      "step": 3405
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.371325792668326e-06,
+      "loss": 2.0906,
+      "step": 3410
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.369520608657122e-06,
+      "loss": 2.1179,
+      "step": 3415
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.367713011025219e-06,
+      "loss": 2.1076,
+      "step": 3420
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.36590300077109e-06,
+      "loss": 2.1312,
+      "step": 3425
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.364090578894547e-06,
+      "loss": 2.0939,
+      "step": 3430
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.362275746396722e-06,
+      "loss": 2.1058,
+      "step": 3435
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.36045850428009e-06,
+      "loss": 2.1112,
+      "step": 3440
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.358638853548455e-06,
+      "loss": 2.0905,
+      "step": 3445
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.356816795206942e-06,
+      "loss": 2.1349,
+      "step": 3450
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.354992330262021e-06,
+      "loss": 2.0989,
+      "step": 3455
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.35316545972148e-06,
+      "loss": 2.109,
+      "step": 3460
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.35133618459444e-06,
+      "loss": 2.0605,
+      "step": 3465
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.34950450589135e-06,
+      "loss": 2.1101,
+      "step": 3470
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.347670424623986e-06,
+      "loss": 2.0987,
+      "step": 3475
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.345833941805451e-06,
+      "loss": 2.1049,
+      "step": 3480
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.343995058450176e-06,
+      "loss": 2.1119,
+      "step": 3485
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.342153775573919e-06,
+      "loss": 2.1118,
+      "step": 3490
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.340310094193757e-06,
+      "loss": 2.1125,
+      "step": 3495
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.338464015328098e-06,
+      "loss": 2.0939,
+      "step": 3500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.336615539996674e-06,
+      "loss": 2.1214,
+      "step": 3505
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.334764669220537e-06,
+      "loss": 2.1163,
+      "step": 3510
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.332911404022068e-06,
+      "loss": 2.1208,
+      "step": 3515
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.331055745424967e-06,
+      "loss": 2.11,
+      "step": 3520
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.32919769445425e-06,
+      "loss": 2.1154,
+      "step": 3525
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.32733725213627e-06,
+      "loss": 2.0846,
+      "step": 3530
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.325474419498684e-06,
+      "loss": 2.1175,
+      "step": 3535
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.323609197570482e-06,
+      "loss": 2.0984,
+      "step": 3540
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.321741587381966e-06,
+      "loss": 2.1171,
+      "step": 3545
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.319871589964763e-06,
+      "loss": 2.0997,
+      "step": 3550
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.317999206351812e-06,
+      "loss": 2.1087,
+      "step": 3555
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.316124437577376e-06,
+      "loss": 2.1119,
+      "step": 3560
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.314247284677033e-06,
+      "loss": 2.121,
+      "step": 3565
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.31236774868768e-06,
+      "loss": 2.0917,
+      "step": 3570
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.310485830647524e-06,
+      "loss": 2.098,
+      "step": 3575
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.308601531596096e-06,
+      "loss": 2.1053,
+      "step": 3580
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.306714852574239e-06,
+      "loss": 2.1038,
+      "step": 3585
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.304825794624108e-06,
+      "loss": 2.125,
+      "step": 3590
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.302934358789175e-06,
+      "loss": 2.0822,
+      "step": 3595
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.301040546114226e-06,
+      "loss": 2.1163,
+      "step": 3600
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.299144357645355e-06,
+      "loss": 2.1005,
+      "step": 3605
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.297245794429978e-06,
+      "loss": 2.0879,
+      "step": 3610
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.295344857516812e-06,
+      "loss": 2.1092,
+      "step": 3615
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.293441547955893e-06,
+      "loss": 2.1219,
+      "step": 3620
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.291535866798563e-06,
+      "loss": 2.0961,
+      "step": 3625
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.289627815097475e-06,
+      "loss": 2.1145,
+      "step": 3630
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.287717393906593e-06,
+      "loss": 2.0836,
+      "step": 3635
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.28580460428119e-06,
+      "loss": 2.0798,
+      "step": 3640
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.283889447277843e-06,
+      "loss": 2.1136,
+      "step": 3645
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.281971923954441e-06,
+      "loss": 2.0902,
+      "step": 3650
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.280052035370181e-06,
+      "loss": 2.1133,
+      "step": 3655
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.278129782585562e-06,
+      "loss": 2.1081,
+      "step": 3660
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.276205166662392e-06,
+      "loss": 2.0988,
+      "step": 3665
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.274278188663784e-06,
+      "loss": 2.1273,
+      "step": 3670
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.272348849654156e-06,
+      "loss": 2.0875,
+      "step": 3675
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.27041715069923e-06,
+      "loss": 2.0931,
+      "step": 3680
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.268483092866027e-06,
+      "loss": 2.1119,
+      "step": 3685
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.266546677222881e-06,
+      "loss": 2.099,
+      "step": 3690
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.26460790483942e-06,
+      "loss": 2.1002,
+      "step": 3695
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.262666776786576e-06,
+      "loss": 2.0929,
+      "step": 3700
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.260723294136582e-06,
+      "loss": 2.1045,
+      "step": 3705
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.258777457962973e-06,
+      "loss": 2.0854,
+      "step": 3710
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.256829269340585e-06,
+      "loss": 2.0839,
+      "step": 3715
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.254878729345549e-06,
+      "loss": 2.1091,
+      "step": 3720
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.252925839055297e-06,
+      "loss": 2.1057,
+      "step": 3725
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.250970599548566e-06,
+      "loss": 2.0982,
+      "step": 3730
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.249013011905376e-06,
+      "loss": 2.0661,
+      "step": 3735
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.247053077207058e-06,
+      "loss": 2.1107,
+      "step": 3740
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.245090796536233e-06,
+      "loss": 2.0809,
+      "step": 3745
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.243126170976817e-06,
+      "loss": 2.075,
+      "step": 3750
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.241159201614025e-06,
+      "loss": 2.1005,
+      "step": 3755
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.239189889534363e-06,
+      "loss": 2.0605,
+      "step": 3760
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.237218235825634e-06,
+      "loss": 2.1096,
+      "step": 3765
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.235244241576935e-06,
+      "loss": 2.0903,
+      "step": 3770
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.233267907878651e-06,
+      "loss": 2.1064,
+      "step": 3775
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.231289235822462e-06,
+      "loss": 2.0695,
+      "step": 3780
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.229308226501343e-06,
+      "loss": 2.1119,
+      "step": 3785
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.227324881009556e-06,
+      "loss": 2.0796,
+      "step": 3790
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.225339200442654e-06,
+      "loss": 2.0977,
+      "step": 3795
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.223351185897482e-06,
+      "loss": 2.0719,
+      "step": 3800
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.221360838472168e-06,
+      "loss": 2.0876,
+      "step": 3805
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.21936815926614e-06,
+      "loss": 2.0981,
+      "step": 3810
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.217373149380102e-06,
+      "loss": 2.083,
+      "step": 3815
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.215375809916051e-06,
+      "loss": 2.0915,
+      "step": 3820
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.213376141977276e-06,
+      "loss": 2.073,
+      "step": 3825
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.211374146668339e-06,
+      "loss": 2.0831,
+      "step": 3830
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.2093698250951e-06,
+      "loss": 2.091,
+      "step": 3835
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.207363178364696e-06,
+      "loss": 2.1082,
+      "step": 3840
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.205354207585557e-06,
+      "loss": 2.077,
+      "step": 3845
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.203342913867386e-06,
+      "loss": 2.1128,
+      "step": 3850
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.201329298321176e-06,
+      "loss": 2.0682,
+      "step": 3855
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.199313362059203e-06,
+      "loss": 2.0896,
+      "step": 3860
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.19729510619502e-06,
+      "loss": 2.1031,
+      "step": 3865
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.195274531843465e-06,
+      "loss": 2.0941,
+      "step": 3870
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.193251640120657e-06,
+      "loss": 2.1051,
+      "step": 3875
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.191226432143994e-06,
+      "loss": 2.1032,
+      "step": 3880
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.189198909032154e-06,
+      "loss": 2.1046,
+      "step": 3885
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.187169071905089e-06,
+      "loss": 2.0806,
+      "step": 3890
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.185136921884037e-06,
+      "loss": 2.0937,
+      "step": 3895
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.183102460091511e-06,
+      "loss": 2.1226,
+      "step": 3900
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.181065687651298e-06,
+      "loss": 2.1033,
+      "step": 3905
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.179026605688464e-06,
+      "loss": 2.0805,
+      "step": 3910
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.176985215329347e-06,
+      "loss": 2.0928,
+      "step": 3915
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.174941517701568e-06,
+      "loss": 2.0882,
+      "step": 3920
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.172895513934016e-06,
+      "loss": 2.092,
+      "step": 3925
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.170847205156854e-06,
+      "loss": 2.1222,
+      "step": 3930
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.16879659250152e-06,
+      "loss": 2.0782,
+      "step": 3935
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.166743677100727e-06,
+      "loss": 2.1024,
+      "step": 3940
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.164688460088455e-06,
+      "loss": 2.1012,
+      "step": 3945
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.162630942599958e-06,
+      "loss": 2.0907,
+      "step": 3950
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.16057112577176e-06,
+      "loss": 2.0972,
+      "step": 3955
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.158509010741657e-06,
+      "loss": 2.0886,
+      "step": 3960
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.15644459864871e-06,
+      "loss": 2.0621,
+      "step": 3965
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.154377890633255e-06,
+      "loss": 2.1059,
+      "step": 3970
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.152308887836892e-06,
+      "loss": 2.0777,
+      "step": 3975
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.150237591402489e-06,
+      "loss": 2.1031,
+      "step": 3980
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.148164002474181e-06,
+      "loss": 2.1204,
+      "step": 3985
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.146088122197372e-06,
+      "loss": 2.1341,
+      "step": 3990
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.144009951718729e-06,
+      "loss": 2.0916,
+      "step": 3995
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.141929492186184e-06,
+      "loss": 2.0959,
+      "step": 4000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.139846744748932e-06,
+      "loss": 2.0698,
+      "step": 4005
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.137761710557438e-06,
+      "loss": 2.1224,
+      "step": 4010
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.135674390763424e-06,
+      "loss": 2.0749,
+      "step": 4015
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.133584786519876e-06,
+      "loss": 2.0806,
+      "step": 4020
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.131492898981043e-06,
+      "loss": 2.1086,
+      "step": 4025
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.129398729302434e-06,
+      "loss": 2.1086,
+      "step": 4030
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.12730227864082e-06,
+      "loss": 2.0861,
+      "step": 4035
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.12520354815423e-06,
+      "loss": 2.0919,
+      "step": 4040
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.123102539001955e-06,
+      "loss": 2.0926,
+      "step": 4045
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.120999252344542e-06,
+      "loss": 2.1094,
+      "step": 4050
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.1188936893438e-06,
+      "loss": 2.0712,
+      "step": 4055
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.116785851162788e-06,
+      "loss": 2.0695,
+      "step": 4060
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.114675738965829e-06,
+      "loss": 2.1158,
+      "step": 4065
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.1125633539185e-06,
+      "loss": 2.0825,
+      "step": 4070
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.11044869718763e-06,
+      "loss": 2.1126,
+      "step": 4075
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.10833176994131e-06,
+      "loss": 2.0793,
+      "step": 4080
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.106212573348878e-06,
+      "loss": 2.0959,
+      "step": 4085
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.10409110858093e-06,
+      "loss": 2.0686,
+      "step": 4090
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.101967376809313e-06,
+      "loss": 2.0775,
+      "step": 4095
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.099841379207126e-06,
+      "loss": 2.0757,
+      "step": 4100
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.09771311694872e-06,
+      "loss": 2.0682,
+      "step": 4105
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.095582591209699e-06,
+      "loss": 2.0666,
+      "step": 4110
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.093449803166914e-06,
+      "loss": 2.0868,
+      "step": 4115
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.091314753998467e-06,
+      "loss": 2.0501,
+      "step": 4120
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.08917744488371e-06,
+      "loss": 2.0699,
+      "step": 4125
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.08703787700324e-06,
+      "loss": 2.1011,
+      "step": 4130
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.084896051538908e-06,
+      "loss": 2.079,
+      "step": 4135
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.082751969673804e-06,
+      "loss": 2.0843,
+      "step": 4140
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.080605632592272e-06,
+      "loss": 2.1012,
+      "step": 4145
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.078457041479895e-06,
+      "loss": 2.089,
+      "step": 4150
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.076306197523507e-06,
+      "loss": 2.0892,
+      "step": 4155
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.074153101911186e-06,
+      "loss": 2.1,
+      "step": 4160
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.071997755832245e-06,
+      "loss": 2.0652,
+      "step": 4165
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.069840160477251e-06,
+      "loss": 2.125,
+      "step": 4170
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.067680317038008e-06,
+      "loss": 2.0755,
+      "step": 4175
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.065518226707564e-06,
+      "loss": 2.0997,
+      "step": 4180
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.063353890680204e-06,
+      "loss": 2.1031,
+      "step": 4185
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.061187310151461e-06,
+      "loss": 2.067,
+      "step": 4190
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.059018486318098e-06,
+      "loss": 2.0671,
+      "step": 4195
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.056847420378128e-06,
+      "loss": 2.0814,
+      "step": 4200
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.054674113530793e-06,
+      "loss": 2.076,
+      "step": 4205
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.052498566976577e-06,
+      "loss": 2.0838,
+      "step": 4210
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.050320781917203e-06,
+      "loss": 2.0859,
+      "step": 4215
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.048140759555626e-06,
+      "loss": 2.0812,
+      "step": 4220
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.04595850109604e-06,
+      "loss": 2.0813,
+      "step": 4225
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.043774007743875e-06,
+      "loss": 2.0974,
+      "step": 4230
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.04158728070579e-06,
+      "loss": 2.0713,
+      "step": 4235
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.039398321189687e-06,
+      "loss": 2.1045,
+      "step": 4240
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.037207130404693e-06,
+      "loss": 2.0904,
+      "step": 4245
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.035013709561168e-06,
+      "loss": 2.0446,
+      "step": 4250
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.032818059870711e-06,
+      "loss": 2.0744,
+      "step": 4255
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.030620182546147e-06,
+      "loss": 2.0847,
+      "step": 4260
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.028420078801526e-06,
+      "loss": 2.0755,
+      "step": 4265
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.02621774985214e-06,
+      "loss": 2.0806,
+      "step": 4270
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.024013196914502e-06,
+      "loss": 2.0803,
+      "step": 4275
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.021806421206354e-06,
+      "loss": 2.0802,
+      "step": 4280
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.019597423946669e-06,
+      "loss": 2.0761,
+      "step": 4285
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.017386206355643e-06,
+      "loss": 2.0582,
+      "step": 4290
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.015172769654703e-06,
+      "loss": 2.093,
+      "step": 4295
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.0129571150665e-06,
+      "loss": 2.0576,
+      "step": 4300
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.010739243814905e-06,
+      "loss": 2.0919,
+      "step": 4305
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.008519157125024e-06,
+      "loss": 2.0897,
+      "step": 4310
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.006296856223178e-06,
+      "loss": 2.0688,
+      "step": 4315
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.004072342336913e-06,
+      "loss": 2.092,
+      "step": 4320
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.001845616695e-06,
+      "loss": 2.0855,
+      "step": 4325
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.99961668052743e-06,
+      "loss": 2.0867,
+      "step": 4330
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.997385535065412e-06,
+      "loss": 2.056,
+      "step": 4335
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.995152181541384e-06,
+      "loss": 2.0873,
+      "step": 4340
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.992916621188992e-06,
+      "loss": 2.0549,
+      "step": 4345
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.990678855243112e-06,
+      "loss": 2.06,
+      "step": 4350
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.988438884939832e-06,
+      "loss": 2.0597,
+      "step": 4355
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.986196711516457e-06,
+      "loss": 2.0648,
+      "step": 4360
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.983952336211513e-06,
+      "loss": 2.0634,
+      "step": 4365
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.98170576026474e-06,
+      "loss": 2.0714,
+      "step": 4370
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.979456984917093e-06,
+      "loss": 2.0688,
+      "step": 4375
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.977206011410744e-06,
+      "loss": 2.0292,
+      "step": 4380
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.97495284098908e-06,
+      "loss": 2.0555,
+      "step": 4385
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.972697474896692e-06,
+      "loss": 2.0364,
+      "step": 4390
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.9704399143794e-06,
+      "loss": 2.0757,
+      "step": 4395
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.968180160684222e-06,
+      "loss": 2.0699,
+      "step": 4400
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.965918215059394e-06,
+      "loss": 2.0642,
+      "step": 4405
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.963654078754365e-06,
+      "loss": 2.089,
+      "step": 4410
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.961387753019786e-06,
+      "loss": 2.0653,
+      "step": 4415
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.959119239107525e-06,
+      "loss": 2.105,
+      "step": 4420
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.956848538270654e-06,
+      "loss": 2.0717,
+      "step": 4425
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.95457565176346e-06,
+      "loss": 2.0469,
+      "step": 4430
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.952300580841425e-06,
+      "loss": 2.0884,
+      "step": 4435
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.950023326761247e-06,
+      "loss": 2.0532,
+      "step": 4440
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.947743890780832e-06,
+      "loss": 2.0745,
+      "step": 4445
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.945462274159281e-06,
+      "loss": 2.0831,
+      "step": 4450
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.943178478156911e-06,
+      "loss": 2.0486,
+      "step": 4455
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.940892504035233e-06,
+      "loss": 2.0706,
+      "step": 4460
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.938604353056968e-06,
+      "loss": 2.1082,
+      "step": 4465
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.936314026486035e-06,
+      "loss": 2.0691,
+      "step": 4470
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.93402152558756e-06,
+      "loss": 2.0609,
+      "step": 4475
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.931726851627863e-06,
+      "loss": 2.0636,
+      "step": 4480
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.929430005874468e-06,
+      "loss": 2.1088,
+      "step": 4485
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.927130989596104e-06,
+      "loss": 2.0697,
+      "step": 4490
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.924829804062687e-06,
+      "loss": 2.089,
+      "step": 4495
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.922526450545342e-06,
+      "loss": 2.0618,
+      "step": 4500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.920220930316387e-06,
+      "loss": 2.0862,
+      "step": 4505
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.917913244649336e-06,
+      "loss": 2.0748,
+      "step": 4510
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.915603394818902e-06,
+      "loss": 2.0622,
+      "step": 4515
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.913291382100988e-06,
+      "loss": 2.0642,
+      "step": 4520
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.910977207772701e-06,
+      "loss": 2.0684,
+      "step": 4525
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.908660873112333e-06,
+      "loss": 2.1004,
+      "step": 4530
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.906342379399375e-06,
+      "loss": 2.0768,
+      "step": 4535
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.90402172791451e-06,
+      "loss": 2.0495,
+      "step": 4540
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.901698919939605e-06,
+      "loss": 2.0698,
+      "step": 4545
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.89937395675773e-06,
+      "loss": 2.0633,
+      "step": 4550
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.897046839653141e-06,
+      "loss": 2.0893,
+      "step": 4555
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.894717569911278e-06,
+      "loss": 2.0886,
+      "step": 4560
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.89238614881878e-06,
+      "loss": 2.0619,
+      "step": 4565
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.890052577663467e-06,
+      "loss": 2.0536,
+      "step": 4570
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.887716857734348e-06,
+      "loss": 2.0714,
+      "step": 4575
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.885378990321621e-06,
+      "loss": 2.0811,
+      "step": 4580
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.88303897671667e-06,
+      "loss": 2.0601,
+      "step": 4585
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.88069681821206e-06,
+      "loss": 2.0622,
+      "step": 4590
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.878352516101549e-06,
+      "loss": 2.0746,
+      "step": 4595
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.876006071680074e-06,
+      "loss": 2.0443,
+      "step": 4600
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.873657486243754e-06,
+      "loss": 2.0549,
+      "step": 4605
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.871306761089893e-06,
+      "loss": 2.099,
+      "step": 4610
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.868953897516976e-06,
+      "loss": 2.0513,
+      "step": 4615
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.86659889682467e-06,
+      "loss": 2.0334,
+      "step": 4620
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.864241760313822e-06,
+      "loss": 2.0627,
+      "step": 4625
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.861882489286459e-06,
+      "loss": 2.0845,
+      "step": 4630
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.859521085045789e-06,
+      "loss": 2.0616,
+      "step": 4635
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.857157548896194e-06,
+      "loss": 2.0606,
+      "step": 4640
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.854791882143234e-06,
+      "loss": 2.0809,
+      "step": 4645
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.852424086093653e-06,
+      "loss": 2.0642,
+      "step": 4650
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.850054162055361e-06,
+      "loss": 2.0818,
+      "step": 4655
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.847682111337454e-06,
+      "loss": 2.0876,
+      "step": 4660
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.845307935250191e-06,
+      "loss": 2.0722,
+      "step": 4665
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.842931635105015e-06,
+      "loss": 2.0661,
+      "step": 4670
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.840553212214538e-06,
+      "loss": 2.0557,
+      "step": 4675
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.838172667892544e-06,
+      "loss": 2.0417,
+      "step": 4680
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.83579000345399e-06,
+      "loss": 2.085,
+      "step": 4685
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.833405220215005e-06,
+      "loss": 2.087,
+      "step": 4690
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.831018319492888e-06,
+      "loss": 2.0851,
+      "step": 4695
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.828629302606105e-06,
+      "loss": 2.0583,
+      "step": 4700
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.826238170874293e-06,
+      "loss": 2.0748,
+      "step": 4705
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.823844925618257e-06,
+      "loss": 2.0572,
+      "step": 4710
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.821449568159973e-06,
+      "loss": 2.0506,
+      "step": 4715
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.819052099822574e-06,
+      "loss": 2.073,
+      "step": 4720
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.81665252193037e-06,
+      "loss": 2.0542,
+      "step": 4725
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.81425083580883e-06,
+      "loss": 2.0794,
+      "step": 4730
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.81184704278459e-06,
+      "loss": 2.0541,
+      "step": 4735
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.80944114418545e-06,
+      "loss": 2.0905,
+      "step": 4740
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.807033141340367e-06,
+      "loss": 2.0981,
+      "step": 4745
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.80462303557947e-06,
+      "loss": 2.0853,
+      "step": 4750
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.802210828234043e-06,
+      "loss": 2.0799,
+      "step": 4755
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.799796520636533e-06,
+      "loss": 2.0605,
+      "step": 4760
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.79738011412055e-06,
+      "loss": 2.0529,
+      "step": 4765
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.794961610020855e-06,
+      "loss": 2.0666,
+      "step": 4770
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.792541009673374e-06,
+      "loss": 2.0626,
+      "step": 4775
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.790118314415193e-06,
+      "loss": 2.073,
+      "step": 4780
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.78769352558455e-06,
+      "loss": 2.0654,
+      "step": 4785
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.785266644520842e-06,
+      "loss": 2.0519,
+      "step": 4790
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.78283767256462e-06,
+      "loss": 2.0605,
+      "step": 4795
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.780406611057593e-06,
+      "loss": 2.0603,
+      "step": 4800
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.77797346134262e-06,
+      "loss": 2.069,
+      "step": 4805
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.775538224763718e-06,
+      "loss": 2.0189,
+      "step": 4810
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.773100902666053e-06,
+      "loss": 2.0651,
+      "step": 4815
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.770661496395946e-06,
+      "loss": 2.0634,
+      "step": 4820
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.768220007300867e-06,
+      "loss": 2.0494,
+      "step": 4825
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.765776436729438e-06,
+      "loss": 2.0922,
+      "step": 4830
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.76333078603143e-06,
+      "loss": 2.0495,
+      "step": 4835
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.76088305655776e-06,
+      "loss": 2.0399,
+      "step": 4840
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.758433249660502e-06,
+      "loss": 2.058,
+      "step": 4845
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.755981366692867e-06,
+      "loss": 2.0805,
+      "step": 4850
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.75352740900922e-06,
+      "loss": 2.0626,
+      "step": 4855
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.75107137796507e-06,
+      "loss": 2.0637,
+      "step": 4860
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.748613274917069e-06,
+      "loss": 2.0423,
+      "step": 4865
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.746153101223016e-06,
+      "loss": 2.0586,
+      "step": 4870
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.743690858241854e-06,
+      "loss": 2.0734,
+      "step": 4875
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.74122654733367e-06,
+      "loss": 2.0769,
+      "step": 4880
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.738760169859688e-06,
+      "loss": 2.0808,
+      "step": 4885
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.736291727182278e-06,
+      "loss": 2.0604,
+      "step": 4890
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.733821220664951e-06,
+      "loss": 2.0591,
+      "step": 4895
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.731348651672357e-06,
+      "loss": 2.057,
+      "step": 4900
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.728874021570284e-06,
+      "loss": 2.0434,
+      "step": 4905
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.72639733172566e-06,
+      "loss": 2.0561,
+      "step": 4910
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.723918583506553e-06,
+      "loss": 2.0436,
+      "step": 4915
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.72143777828216e-06,
+      "loss": 2.0589,
+      "step": 4920
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.718954917422823e-06,
+      "loss": 2.0339,
+      "step": 4925
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.716470002300015e-06,
+      "loss": 2.0637,
+      "step": 4930
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.713983034286347e-06,
+      "loss": 2.0772,
+      "step": 4935
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.71149401475556e-06,
+      "loss": 2.0871,
+      "step": 4940
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.70900294508253e-06,
+      "loss": 2.0484,
+      "step": 4945
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.706509826643265e-06,
+      "loss": 2.0818,
+      "step": 4950
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.704014660814905e-06,
+      "loss": 2.0621,
+      "step": 4955
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.701517448975725e-06,
+      "loss": 2.0458,
+      "step": 4960
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.69901819250512e-06,
+      "loss": 2.0614,
+      "step": 4965
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.696516892783627e-06,
+      "loss": 2.0555,
+      "step": 4970
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.6940135511929e-06,
+      "loss": 2.0372,
+      "step": 4975
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.691508169115728e-06,
+      "loss": 2.0589,
+      "step": 4980
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.689000747936028e-06,
+      "loss": 2.038,
+      "step": 4985
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.686491289038835e-06,
+      "loss": 2.0404,
+      "step": 4990
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.68397979381032e-06,
+      "loss": 2.0814,
+      "step": 4995
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.681466263637773e-06,
+      "loss": 2.0499,
+      "step": 5000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.67895069990961e-06,
+      "loss": 2.0565,
+      "step": 5005
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.676433104015365e-06,
+      "loss": 2.0598,
+      "step": 5010
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.673913477345703e-06,
+      "loss": 2.0745,
+      "step": 5015
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.671391821292407e-06,
+      "loss": 2.0702,
+      "step": 5020
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.668868137248379e-06,
+      "loss": 2.0428,
+      "step": 5025
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.666342426607643e-06,
+      "loss": 2.0665,
+      "step": 5030
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.663814690765341e-06,
+      "loss": 2.0545,
+      "step": 5035
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.66128493111774e-06,
+      "loss": 2.0635,
+      "step": 5040
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.658753149062214e-06,
+      "loss": 2.0773,
+      "step": 5045
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.656219345997262e-06,
+      "loss": 2.0638,
+      "step": 5050
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.653683523322498e-06,
+      "loss": 2.0613,
+      "step": 5055
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.65114568243865e-06,
+      "loss": 2.0552,
+      "step": 5060
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.648605824747565e-06,
+      "loss": 2.0524,
+      "step": 5065
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.646063951652198e-06,
+      "loss": 2.0539,
+      "step": 5070
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.64352006455662e-06,
+      "loss": 2.0473,
+      "step": 5075
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.640974164866015e-06,
+      "loss": 2.0337,
+      "step": 5080
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.638426253986678e-06,
+      "loss": 2.0766,
+      "step": 5085
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.635876333326017e-06,
+      "loss": 2.0422,
+      "step": 5090
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.633324404292544e-06,
+      "loss": 2.0417,
+      "step": 5095
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.63077046829589e-06,
+      "loss": 2.0305,
+      "step": 5100
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.628214526746786e-06,
+      "loss": 2.0408,
+      "step": 5105
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.625656581057076e-06,
+      "loss": 2.0415,
+      "step": 5110
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.623096632639707e-06,
+      "loss": 2.0669,
+      "step": 5115
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.620534682908736e-06,
+      "loss": 2.0635,
+      "step": 5120
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.617970733279325e-06,
+      "loss": 2.0966,
+      "step": 5125
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.615404785167738e-06,
+      "loss": 2.0467,
+      "step": 5130
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.612836839991346e-06,
+      "loss": 2.0599,
+      "step": 5135
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.610266899168618e-06,
+      "loss": 2.0451,
+      "step": 5140
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.607694964119132e-06,
+      "loss": 2.0679,
+      "step": 5145
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.605121036263567e-06,
+      "loss": 2.0541,
+      "step": 5150
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.602545117023694e-06,
+      "loss": 2.0797,
+      "step": 5155
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.599967207822397e-06,
+      "loss": 2.0511,
+      "step": 5160
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.597387310083649e-06,
+      "loss": 2.0538,
+      "step": 5165
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.594805425232525e-06,
+      "loss": 2.0432,
+      "step": 5170
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.592221554695199e-06,
+      "loss": 2.0336,
+      "step": 5175
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.589635699898939e-06,
+      "loss": 2.0464,
+      "step": 5180
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.587047862272111e-06,
+      "loss": 2.058,
+      "step": 5185
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.584458043244177e-06,
+      "loss": 2.0511,
+      "step": 5190
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.581866244245691e-06,
+      "loss": 2.0658,
+      "step": 5195
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.5792724667083e-06,
+      "loss": 2.0383,
+      "step": 5200
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.576676712064752e-06,
+      "loss": 2.0582,
+      "step": 5205
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.574078981748875e-06,
+      "loss": 2.069,
+      "step": 5210
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.571479277195598e-06,
+      "loss": 2.0385,
+      "step": 5215
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.568877599840934e-06,
+      "loss": 2.0579,
+      "step": 5220
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.566273951121991e-06,
+      "loss": 2.0439,
+      "step": 5225
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.56366833247696e-06,
+      "loss": 2.0672,
+      "step": 5230
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.561060745345128e-06,
+      "loss": 2.0652,
+      "step": 5235
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.558451191166863e-06,
+      "loss": 2.0587,
+      "step": 5240
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.555839671383622e-06,
+      "loss": 2.0384,
+      "step": 5245
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.553226187437945e-06,
+      "loss": 2.0585,
+      "step": 5250
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.550610740773462e-06,
+      "loss": 2.046,
+      "step": 5255
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.547993332834882e-06,
+      "loss": 2.0749,
+      "step": 5260
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.545373965068002e-06,
+      "loss": 2.0577,
+      "step": 5265
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.5427526389197e-06,
+      "loss": 2.0572,
+      "step": 5270
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.540129355837932e-06,
+      "loss": 2.0409,
+      "step": 5275
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.537504117271739e-06,
+      "loss": 2.0485,
+      "step": 5280
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.534876924671243e-06,
+      "loss": 2.0698,
+      "step": 5285
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.532247779487641e-06,
+      "loss": 2.0051,
+      "step": 5290
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.529616683173215e-06,
+      "loss": 2.0781,
+      "step": 5295
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.526983637181316e-06,
+      "loss": 2.0542,
+      "step": 5300
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.52434864296638e-06,
+      "loss": 2.0389,
+      "step": 5305
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.521711701983915e-06,
+      "loss": 2.0373,
+      "step": 5310
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.519072815690505e-06,
+      "loss": 2.0637,
+      "step": 5315
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.516431985543812e-06,
+      "loss": 2.0545,
+      "step": 5320
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.513789213002563e-06,
+      "loss": 2.0386,
+      "step": 5325
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.511144499526568e-06,
+      "loss": 2.0461,
+      "step": 5330
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.508497846576703e-06,
+      "loss": 2.0334,
+      "step": 5335
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.505849255614918e-06,
+      "loss": 2.0668,
+      "step": 5340
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.503198728104232e-06,
+      "loss": 2.0538,
+      "step": 5345
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.500546265508734e-06,
+      "loss": 2.063,
+      "step": 5350
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.497891869293582e-06,
+      "loss": 2.0095,
+      "step": 5355
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.495235540925003e-06,
+      "loss": 2.0184,
+      "step": 5360
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.492577281870289e-06,
+      "loss": 2.0665,
+      "step": 5365
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.489917093597801e-06,
+      "loss": 2.0518,
+      "step": 5370
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.487254977576967e-06,
+      "loss": 2.0641,
+      "step": 5375
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.484590935278272e-06,
+      "loss": 2.0332,
+      "step": 5380
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.481924968173275e-06,
+      "loss": 2.0431,
+      "step": 5385
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.479257077734591e-06,
+      "loss": 2.0458,
+      "step": 5390
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.476587265435903e-06,
+      "loss": 2.0528,
+      "step": 5395
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.473915532751948e-06,
+      "loss": 2.0533,
+      "step": 5400
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.471241881158534e-06,
+      "loss": 2.0647,
+      "step": 5405
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.46856631213252e-06,
+      "loss": 2.0431,
+      "step": 5410
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.465888827151829e-06,
+      "loss": 2.0194,
+      "step": 5415
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.463209427695441e-06,
+      "loss": 2.0102,
+      "step": 5420
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.460528115243391e-06,
+      "loss": 2.0306,
+      "step": 5425
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.457844891276776e-06,
+      "loss": 2.0814,
+      "step": 5430
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.455159757277744e-06,
+      "loss": 2.045,
+      "step": 5435
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.452472714729502e-06,
+      "loss": 2.0501,
+      "step": 5440
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.449783765116308e-06,
+      "loss": 2.0375,
+      "step": 5445
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.447092909923476e-06,
+      "loss": 2.0104,
+      "step": 5450
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.444400150637368e-06,
+      "loss": 2.0707,
+      "step": 5455
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.441705488745405e-06,
+      "loss": 2.0427,
+      "step": 5460
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.439008925736049e-06,
+      "loss": 2.0417,
+      "step": 5465
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.436310463098826e-06,
+      "loss": 2.016,
+      "step": 5470
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.433610102324301e-06,
+      "loss": 2.0452,
+      "step": 5475
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.430907844904088e-06,
+      "loss": 2.0406,
+      "step": 5480
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.42820369233085e-06,
+      "loss": 2.0455,
+      "step": 5485
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.4254976460983e-06,
+      "loss": 2.0535,
+      "step": 5490
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.422789707701193e-06,
+      "loss": 2.043,
+      "step": 5495
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.420079878635332e-06,
+      "loss": 2.0056,
+      "step": 5500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.417368160397561e-06,
+      "loss": 2.0326,
+      "step": 5505
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.41465455448577e-06,
+      "loss": 2.0414,
+      "step": 5510
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.411939062398894e-06,
+      "loss": 2.0407,
+      "step": 5515
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.409221685636904e-06,
+      "loss": 2.0485,
+      "step": 5520
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.406502425700816e-06,
+      "loss": 2.0191,
+      "step": 5525
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.403781284092688e-06,
+      "loss": 2.055,
+      "step": 5530
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.401058262315609e-06,
+      "loss": 2.0153,
+      "step": 5535
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.398333361873716e-06,
+      "loss": 2.0574,
+      "step": 5540
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.395606584272183e-06,
+      "loss": 2.0609,
+      "step": 5545
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.392877931017212e-06,
+      "loss": 2.0334,
+      "step": 5550
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.390147403616051e-06,
+      "loss": 2.059,
+      "step": 5555
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.387415003576978e-06,
+      "loss": 2.0451,
+      "step": 5560
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.384680732409308e-06,
+      "loss": 2.0667,
+      "step": 5565
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.381944591623385e-06,
+      "loss": 2.0579,
+      "step": 5570
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.379206582730593e-06,
+      "loss": 2.0442,
+      "step": 5575
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.37646670724334e-06,
+      "loss": 2.0121,
+      "step": 5580
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.373724966675074e-06,
+      "loss": 2.0533,
+      "step": 5585
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.370981362540264e-06,
+      "loss": 1.9998,
+      "step": 5590
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.368235896354413e-06,
+      "loss": 2.0668,
+      "step": 5595
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.365488569634055e-06,
+      "loss": 2.0371,
+      "step": 5600
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.362739383896743e-06,
+      "loss": 2.044,
+      "step": 5605
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.35998834066107e-06,
+      "loss": 2.0401,
+      "step": 5610
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.357235441446643e-06,
+      "loss": 2.0573,
+      "step": 5615
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.3544806877741e-06,
+      "loss": 2.0603,
+      "step": 5620
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.351724081165102e-06,
+      "loss": 2.043,
+      "step": 5625
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.348965623142336e-06,
+      "loss": 2.0401,
+      "step": 5630
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.346205315229507e-06,
+      "loss": 2.0762,
+      "step": 5635
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.343443158951345e-06,
+      "loss": 2.0424,
+      "step": 5640
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.3406791558336e-06,
+      "loss": 2.0195,
+      "step": 5645
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.337913307403045e-06,
+      "loss": 2.0297,
+      "step": 5650
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.335145615187467e-06,
+      "loss": 2.0416,
+      "step": 5655
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.332376080715675e-06,
+      "loss": 2.0272,
+      "step": 5660
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.329604705517494e-06,
+      "loss": 2.0343,
+      "step": 5665
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.32683149112377e-06,
+      "loss": 2.0565,
+      "step": 5670
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.324056439066357e-06,
+      "loss": 2.0461,
+      "step": 5675
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.321279550878132e-06,
+      "loss": 2.0504,
+      "step": 5680
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.31850082809298e-06,
+      "loss": 2.0376,
+      "step": 5685
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.315720272245807e-06,
+      "loss": 2.0582,
+      "step": 5690
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.312937884872522e-06,
+      "loss": 2.0402,
+      "step": 5695
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.310153667510054e-06,
+      "loss": 2.0427,
+      "step": 5700
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.307367621696336e-06,
+      "loss": 2.043,
+      "step": 5705
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.304579748970318e-06,
+      "loss": 2.0209,
+      "step": 5710
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.301790050871952e-06,
+      "loss": 2.0175,
+      "step": 5715
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.298998528942207e-06,
+      "loss": 2.0541,
+      "step": 5720
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.296205184723048e-06,
+      "loss": 2.0451,
+      "step": 5725
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.293410019757453e-06,
+      "loss": 2.0034,
+      "step": 5730
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.290613035589413e-06,
+      "loss": 2.0605,
+      "step": 5735
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.28781423376391e-06,
+      "loss": 2.0305,
+      "step": 5740
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.285013615826936e-06,
+      "loss": 2.0467,
+      "step": 5745
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.28221118332549e-06,
+      "loss": 2.0304,
+      "step": 5750
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.27940693780757e-06,
+      "loss": 2.0355,
+      "step": 5755
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.276600880822174e-06,
+      "loss": 2.0519,
+      "step": 5760
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.273793013919302e-06,
+      "loss": 2.0248,
+      "step": 5765
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.270983338649953e-06,
+      "loss": 2.0256,
+      "step": 5770
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.268171856566129e-06,
+      "loss": 2.0415,
+      "step": 5775
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.265358569220822e-06,
+      "loss": 2.0292,
+      "step": 5780
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.26254347816803e-06,
+      "loss": 2.0375,
+      "step": 5785
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.259726584962742e-06,
+      "loss": 2.0625,
+      "step": 5790
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.256907891160941e-06,
+      "loss": 2.0195,
+      "step": 5795
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.254087398319612e-06,
+      "loss": 2.0457,
+      "step": 5800
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.251265107996727e-06,
+      "loss": 2.0494,
+      "step": 5805
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.248441021751251e-06,
+      "loss": 2.0592,
+      "step": 5810
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.245615141143149e-06,
+      "loss": 2.0398,
+      "step": 5815
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.242787467733363e-06,
+      "loss": 2.0222,
+      "step": 5820
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.239958003083841e-06,
+      "loss": 2.0315,
+      "step": 5825
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.237126748757507e-06,
+      "loss": 2.0317,
+      "step": 5830
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.234293706318284e-06,
+      "loss": 2.0218,
+      "step": 5835
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.231458877331074e-06,
+      "loss": 2.0176,
+      "step": 5840
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.228622263361772e-06,
+      "loss": 2.0758,
+      "step": 5845
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.22578386597726e-06,
+      "loss": 2.0611,
+      "step": 5850
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.222943686745397e-06,
+      "loss": 2.015,
+      "step": 5855
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.220101727235035e-06,
+      "loss": 2.0179,
+      "step": 5860
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.217257989016004e-06,
+      "loss": 2.0235,
+      "step": 5865
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.21441247365912e-06,
+      "loss": 2.0279,
+      "step": 5870
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.211565182736176e-06,
+      "loss": 2.0553,
+      "step": 5875
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.20871611781995e-06,
+      "loss": 2.0328,
+      "step": 5880
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.205865280484201e-06,
+      "loss": 2.0258,
+      "step": 5885
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.203012672303662e-06,
+      "loss": 2.04,
+      "step": 5890
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.200158294854048e-06,
+      "loss": 2.014,
+      "step": 5895
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.19730214971205e-06,
+      "loss": 2.0861,
+      "step": 5900
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.194444238455334e-06,
+      "loss": 2.0559,
+      "step": 5905
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.191584562662546e-06,
+      "loss": 2.0278,
+      "step": 5910
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.188723123913303e-06,
+      "loss": 2.0195,
+      "step": 5915
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.185859923788195e-06,
+      "loss": 2.0377,
+      "step": 5920
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.182994963868787e-06,
+      "loss": 2.0138,
+      "step": 5925
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.180128245737617e-06,
+      "loss": 2.0173,
+      "step": 5930
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.177259770978193e-06,
+      "loss": 2.0301,
+      "step": 5935
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.174389541174992e-06,
+      "loss": 2.0443,
+      "step": 5940
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.17151755791346e-06,
+      "loss": 2.0467,
+      "step": 5945
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.168643822780017e-06,
+      "loss": 2.0418,
+      "step": 5950
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.165768337362045e-06,
+      "loss": 2.0413,
+      "step": 5955
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.162891103247893e-06,
+      "loss": 2.0406,
+      "step": 5960
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.160012122026883e-06,
+      "loss": 2.0575,
+      "step": 5965
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.15713139528929e-06,
+      "loss": 1.9951,
+      "step": 5970
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.154248924626365e-06,
+      "loss": 2.0375,
+      "step": 5975
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.151364711630314e-06,
+      "loss": 2.0278,
+      "step": 5980
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.14847875789431e-06,
+      "loss": 2.0177,
+      "step": 5985
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.145591065012487e-06,
+      "loss": 2.0334,
+      "step": 5990
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.142701634579935e-06,
+      "loss": 2.0369,
+      "step": 5995
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.139810468192713e-06,
+      "loss": 2.025,
+      "step": 6000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.136917567447829e-06,
+      "loss": 2.0164,
+      "step": 6005
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.134022933943255e-06,
+      "loss": 2.0338,
+      "step": 6010
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.13112656927792e-06,
+      "loss": 2.0348,
+      "step": 6015
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.128228475051708e-06,
+      "loss": 2.0202,
+      "step": 6020
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.125328652865457e-06,
+      "loss": 2.0258,
+      "step": 6025
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.12242710432096e-06,
+      "loss": 2.0189,
+      "step": 6030
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.119523831020965e-06,
+      "loss": 2.0085,
+      "step": 6035
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.116618834569175e-06,
+      "loss": 2.0118,
+      "step": 6040
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.113712116570236e-06,
+      "loss": 2.0579,
+      "step": 6045
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.110803678629757e-06,
+      "loss": 2.0191,
+      "step": 6050
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.107893522354285e-06,
+      "loss": 2.0233,
+      "step": 6055
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.104981649351327e-06,
+      "loss": 2.0239,
+      "step": 6060
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.10206806122933e-06,
+      "loss": 2.024,
+      "step": 6065
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.099152759597698e-06,
+      "loss": 2.0503,
+      "step": 6070
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.096235746066765e-06,
+      "loss": 2.032,
+      "step": 6075
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.09331702224783e-06,
+      "loss": 2.0138,
+      "step": 6080
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.090396589753122e-06,
+      "loss": 2.0257,
+      "step": 6085
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.08747445019582e-06,
+      "loss": 2.0231,
+      "step": 6090
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.084550605190049e-06,
+      "loss": 2.0357,
+      "step": 6095
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.08162505635087e-06,
+      "loss": 2.0387,
+      "step": 6100
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.078697805294287e-06,
+      "loss": 2.0072,
+      "step": 6105
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.075768853637245e-06,
+      "loss": 2.0431,
+      "step": 6110
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.07283820299763e-06,
+      "loss": 2.0284,
+      "step": 6115
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.069905854994262e-06,
+      "loss": 2.0229,
+      "step": 6120
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.066971811246903e-06,
+      "loss": 2.0027,
+      "step": 6125
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.06403607337625e-06,
+      "loss": 2.0107,
+      "step": 6130
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.061098643003935e-06,
+      "loss": 2.0038,
+      "step": 6135
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.058159521752527e-06,
+      "loss": 2.0333,
+      "step": 6140
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.055218711245525e-06,
+      "loss": 2.0332,
+      "step": 6145
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.052276213107366e-06,
+      "loss": 2.0346,
+      "step": 6150
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.049332028963414e-06,
+      "loss": 2.0238,
+      "step": 6155
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.046386160439975e-06,
+      "loss": 2.0092,
+      "step": 6160
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.043438609164266e-06,
+      "loss": 2.0232,
+      "step": 6165
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.040489376764455e-06,
+      "loss": 2.0001,
+      "step": 6170
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.037538464869624e-06,
+      "loss": 2.0088,
+      "step": 6175
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.03458587510979e-06,
+      "loss": 2.036,
+      "step": 6180
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.031631609115895e-06,
+      "loss": 2.0263,
+      "step": 6185
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.028675668519801e-06,
+      "loss": 2.0366,
+      "step": 6190
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.025718054954303e-06,
+      "loss": 2.0408,
+      "step": 6195
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.022758770053121e-06,
+      "loss": 2.0511,
+      "step": 6200
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.01979781545089e-06,
+      "loss": 2.0235,
+      "step": 6205
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.016835192783172e-06,
+      "loss": 2.0351,
+      "step": 6210
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.013870903686454e-06,
+      "loss": 2.024,
+      "step": 6215
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.010904949798138e-06,
+      "loss": 2.0262,
+      "step": 6220
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.007937332756544e-06,
+      "loss": 2.0373,
+      "step": 6225
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.004968054200916e-06,
+      "loss": 1.9999,
+      "step": 6230
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.001997115771419e-06,
+      "loss": 1.9956,
+      "step": 6235
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.999024519109122e-06,
+      "loss": 2.0125,
+      "step": 6240
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.996050265856023e-06,
+      "loss": 1.9962,
+      "step": 6245
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.993074357655027e-06,
+      "loss": 2.036,
+      "step": 6250
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.990096796149958e-06,
+      "loss": 2.0602,
+      "step": 6255
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.987117582985547e-06,
+      "loss": 2.0017,
+      "step": 6260
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.984136719807445e-06,
+      "loss": 1.9898,
+      "step": 6265
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.981154208262212e-06,
+      "loss": 2.0357,
+      "step": 6270
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.978170049997316e-06,
+      "loss": 2.0214,
+      "step": 6275
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.975184246661135e-06,
+      "loss": 2.0231,
+      "step": 6280
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.972196799902955e-06,
+      "loss": 2.0103,
+      "step": 6285
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.969207711372978e-06,
+      "loss": 2.0492,
+      "step": 6290
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.9662169827223e-06,
+      "loss": 2.0387,
+      "step": 6295
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.963224615602932e-06,
+      "loss": 2.043,
+      "step": 6300
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.960230611667784e-06,
+      "loss": 2.0492,
+      "step": 6305
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.957234972570683e-06,
+      "loss": 2.0138,
+      "step": 6310
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.954237699966339e-06,
+      "loss": 2.0108,
+      "step": 6315
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.951238795510381e-06,
+      "loss": 2.0082,
+      "step": 6320
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.948238260859331e-06,
+      "loss": 2.0308,
+      "step": 6325
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.945236097670615e-06,
+      "loss": 2.0462,
+      "step": 6330
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.942232307602558e-06,
+      "loss": 2.0191,
+      "step": 6335
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.939226892314383e-06,
+      "loss": 2.0257,
+      "step": 6340
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.936219853466212e-06,
+      "loss": 2.052,
+      "step": 6345
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.93321119271906e-06,
+      "loss": 2.0218,
+      "step": 6350
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.93020091173484e-06,
+      "loss": 2.0297,
+      "step": 6355
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.927189012176365e-06,
+      "loss": 2.0138,
+      "step": 6360
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.924175495707334e-06,
+      "loss": 2.0104,
+      "step": 6365
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.921160363992345e-06,
+      "loss": 2.0157,
+      "step": 6370
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.918143618696884e-06,
+      "loss": 2.0233,
+      "step": 6375
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.915125261487333e-06,
+      "loss": 2.0393,
+      "step": 6380
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.912105294030959e-06,
+      "loss": 2.0162,
+      "step": 6385
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.909083717995923e-06,
+      "loss": 2.0416,
+      "step": 6390
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.906060535051271e-06,
+      "loss": 2.0293,
+      "step": 6395
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.90303574686694e-06,
+      "loss": 2.0324,
+      "step": 6400
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.90000935511375e-06,
+      "loss": 2.0328,
+      "step": 6405
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.896981361463415e-06,
+      "loss": 2.0149,
+      "step": 6410
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.89395176758852e-06,
+      "loss": 1.9871,
+      "step": 6415
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.890920575162545e-06,
+      "loss": 2.0156,
+      "step": 6420
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.887887785859847e-06,
+      "loss": 2.0387,
+      "step": 6425
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.884853401355672e-06,
+      "loss": 2.0142,
+      "step": 6430
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.88181742332614e-06,
+      "loss": 2.0051,
+      "step": 6435
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.878779853448255e-06,
+      "loss": 2.0058,
+      "step": 6440
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.875740693399898e-06,
+      "loss": 2.0247,
+      "step": 6445
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.872699944859832e-06,
+      "loss": 2.0515,
+      "step": 6450
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.869657609507695e-06,
+      "loss": 2.038,
+      "step": 6455
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.866613689024e-06,
+      "loss": 2.0316,
+      "step": 6460
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.863568185090138e-06,
+      "loss": 2.0211,
+      "step": 6465
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.860521099388372e-06,
+      "loss": 2.0355,
+      "step": 6470
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.857472433601845e-06,
+      "loss": 1.9977,
+      "step": 6475
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.854422189414566e-06,
+      "loss": 2.0206,
+      "step": 6480
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.851370368511418e-06,
+      "loss": 2.0336,
+      "step": 6485
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.848316972578156e-06,
+      "loss": 2.0041,
+      "step": 6490
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.845262003301404e-06,
+      "loss": 2.0334,
+      "step": 6495
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.842205462368653e-06,
+      "loss": 2.0365,
+      "step": 6500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.83914735146827e-06,
+      "loss": 2.0315,
+      "step": 6505
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.836087672289478e-06,
+      "loss": 2.0232,
+      "step": 6510
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.833026426522377e-06,
+      "loss": 2.0056,
+      "step": 6515
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.829963615857925e-06,
+      "loss": 2.0304,
+      "step": 6520
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.826899241987947e-06,
+      "loss": 2.0128,
+      "step": 6525
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.823833306605129e-06,
+      "loss": 2.0319,
+      "step": 6530
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.820765811403024e-06,
+      "loss": 2.0179,
+      "step": 6535
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.817696758076045e-06,
+      "loss": 2.031,
+      "step": 6540
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.814626148319464e-06,
+      "loss": 2.0185,
+      "step": 6545
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.811553983829417e-06,
+      "loss": 2.0182,
+      "step": 6550
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.80848026630289e-06,
+      "loss": 2.0424,
+      "step": 6555
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.805404997437736e-06,
+      "loss": 2.0539,
+      "step": 6560
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.802328178932659e-06,
+      "loss": 2.0339,
+      "step": 6565
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.799249812487224e-06,
+      "loss": 2.0264,
+      "step": 6570
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.796169899801848e-06,
+      "loss": 2.0661,
+      "step": 6575
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.7930884425778e-06,
+      "loss": 2.0264,
+      "step": 6580
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.79000544251721e-06,
+      "loss": 1.9978,
+      "step": 6585
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.78692090132305e-06,
+      "loss": 2.0275,
+      "step": 6590
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.783834820699148e-06,
+      "loss": 2.013,
+      "step": 6595
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.780747202350188e-06,
+      "loss": 2.0112,
+      "step": 6600
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.77765804798169e-06,
+      "loss": 2.0137,
+      "step": 6605
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.774567359300038e-06,
+      "loss": 2.0249,
+      "step": 6610
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.771475138012452e-06,
+      "loss": 2.0185,
+      "step": 6615
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.768381385827002e-06,
+      "loss": 2.0136,
+      "step": 6620
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.765286104452605e-06,
+      "loss": 2.0162,
+      "step": 6625
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.762189295599023e-06,
+      "loss": 2.0148,
+      "step": 6630
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.75909096097686e-06,
+      "loss": 2.0398,
+      "step": 6635
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.755991102297562e-06,
+      "loss": 2.0125,
+      "step": 6640
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.75288972127342e-06,
+      "loss": 2.0035,
+      "step": 6645
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.749786819617564e-06,
+      "loss": 2.019,
+      "step": 6650
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.746682399043963e-06,
+      "loss": 2.0195,
+      "step": 6655
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.743576461267427e-06,
+      "loss": 2.0209,
+      "step": 6660
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.740469008003602e-06,
+      "loss": 2.0257,
+      "step": 6665
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.737360040968975e-06,
+      "loss": 2.0157,
+      "step": 6670
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.734249561880863e-06,
+      "loss": 2.0072,
+      "step": 6675
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.731137572457423e-06,
+      "loss": 2.0187,
+      "step": 6680
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.728024074417647e-06,
+      "loss": 2.0111,
+      "step": 6685
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.724909069481355e-06,
+      "loss": 2.0281,
+      "step": 6690
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.721792559369204e-06,
+      "loss": 2.0009,
+      "step": 6695
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.718674545802683e-06,
+      "loss": 2.0131,
+      "step": 6700
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.715555030504107e-06,
+      "loss": 2.0198,
+      "step": 6705
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.712434015196622e-06,
+      "loss": 2.0419,
+      "step": 6710
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.709311501604208e-06,
+      "loss": 2.0169,
+      "step": 6715
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.706187491451667e-06,
+      "loss": 2.0225,
+      "step": 6720
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.703061986464625e-06,
+      "loss": 2.0105,
+      "step": 6725
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.699934988369541e-06,
+      "loss": 1.9857,
+      "step": 6730
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.696806498893697e-06,
+      "loss": 2.0014,
+      "step": 6735
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.693676519765193e-06,
+      "loss": 2.0179,
+      "step": 6740
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.69054505271296e-06,
+      "loss": 2.0391,
+      "step": 6745
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.687412099466742e-06,
+      "loss": 2.0245,
+      "step": 6750
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.684277661757113e-06,
+      "loss": 2.0179,
+      "step": 6755
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.681141741315464e-06,
+      "loss": 2.0119,
+      "step": 6760
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.678004339874e-06,
+      "loss": 2.0274,
+      "step": 6765
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.67486545916575e-06,
+      "loss": 2.0418,
+      "step": 6770
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.671725100924558e-06,
+      "loss": 2.0231,
+      "step": 6775
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.668583266885084e-06,
+      "loss": 2.008,
+      "step": 6780
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.665439958782803e-06,
+      "loss": 2.0072,
+      "step": 6785
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.662295178354006e-06,
+      "loss": 2.0157,
+      "step": 6790
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.659148927335796e-06,
+      "loss": 2.033,
+      "step": 6795
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.656001207466086e-06,
+      "loss": 2.0302,
+      "step": 6800
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.652852020483604e-06,
+      "loss": 2.0277,
+      "step": 6805
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.649701368127886e-06,
+      "loss": 2.0308,
+      "step": 6810
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.646549252139282e-06,
+      "loss": 2.0065,
+      "step": 6815
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.643395674258943e-06,
+      "loss": 2.0063,
+      "step": 6820
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.640240636228833e-06,
+      "loss": 2.0059,
+      "step": 6825
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.63708413979172e-06,
+      "loss": 2.0303,
+      "step": 6830
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.63392618669118e-06,
+      "loss": 2.0256,
+      "step": 6835
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.630766778671595e-06,
+      "loss": 2.0587,
+      "step": 6840
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.6276059174781406e-06,
+      "loss": 2.0085,
+      "step": 6845
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.624443604856807e-06,
+      "loss": 2.0351,
+      "step": 6850
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.621279842554382e-06,
+      "loss": 2.0291,
+      "step": 6855
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.618114632318452e-06,
+      "loss": 2.0096,
+      "step": 6860
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.614947975897407e-06,
+      "loss": 2.0335,
+      "step": 6865
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.61177987504043e-06,
+      "loss": 2.0143,
+      "step": 6870
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.608610331497509e-06,
+      "loss": 2.0127,
+      "step": 6875
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.605439347019423e-06,
+      "loss": 2.0046,
+      "step": 6880
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.6022669233577524e-06,
+      "loss": 2.0286,
+      "step": 6885
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.599093062264868e-06,
+      "loss": 2.0134,
+      "step": 6890
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.595917765493935e-06,
+      "loss": 2.0173,
+      "step": 6895
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.592741034798916e-06,
+      "loss": 2.021,
+      "step": 6900
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.589562871934559e-06,
+      "loss": 1.9705,
+      "step": 6905
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.5863832786564095e-06,
+      "loss": 2.0348,
+      "step": 6910
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.583202256720798e-06,
+      "loss": 2.0242,
+      "step": 6915
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.5800198078848485e-06,
+      "loss": 2.0328,
+      "step": 6920
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.576835933906469e-06,
+      "loss": 2.0154,
+      "step": 6925
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.5736506365443585e-06,
+      "loss": 2.0246,
+      "step": 6930
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.570463917558e-06,
+      "loss": 2.0009,
+      "step": 6935
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.567275778707661e-06,
+      "loss": 2.0071,
+      "step": 6940
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.564086221754399e-06,
+      "loss": 2.0178,
+      "step": 6945
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.560895248460043e-06,
+      "loss": 2.0152,
+      "step": 6950
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.5577028605872194e-06,
+      "loss": 2.0119,
+      "step": 6955
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.554509059899322e-06,
+      "loss": 2.0087,
+      "step": 6960
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.551313848160538e-06,
+      "loss": 2.0293,
+      "step": 6965
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.548117227135821e-06,
+      "loss": 2.0114,
+      "step": 6970
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.544919198590915e-06,
+      "loss": 1.9963,
+      "step": 6975
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.541719764292331e-06,
+      "loss": 2.0151,
+      "step": 6980
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.538518926007364e-06,
+      "loss": 2.0124,
+      "step": 6985
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.535316685504084e-06,
+      "loss": 2.0023,
+      "step": 6990
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.532113044551329e-06,
+      "loss": 2.0072,
+      "step": 6995
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.528908004918719e-06,
+      "loss": 1.981,
+      "step": 7000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.5257015683766395e-06,
+      "loss": 2.0227,
+      "step": 7005
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.522493736696255e-06,
+      "loss": 2.0098,
+      "step": 7010
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.519284511649492e-06,
+      "loss": 2.0101,
+      "step": 7015
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.516073895009053e-06,
+      "loss": 1.984,
+      "step": 7020
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.5128618885484084e-06,
+      "loss": 2.0045,
+      "step": 7025
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.509648494041794e-06,
+      "loss": 1.9996,
+      "step": 7030
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.5064337132642154e-06,
+      "loss": 1.9918,
+      "step": 7035
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.503217547991439e-06,
+      "loss": 2.0126,
+      "step": 7040
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 2.0137,
+      "step": 7045
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.4967810710672e-06,
+      "loss": 2.0081,
+      "step": 7050
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.493560762971094e-06,
+      "loss": 2.0387,
+      "step": 7055
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.490339077490509e-06,
+      "loss": 2.0102,
+      "step": 7060
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.487116016405026e-06,
+      "loss": 2.0204,
+      "step": 7065
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.483891581494991e-06,
+      "loss": 1.9957,
+      "step": 7070
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.480665774541502e-06,
+      "loss": 2.0241,
+      "step": 7075
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.477438597326424e-06,
+      "loss": 1.9892,
+      "step": 7080
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.474210051632372e-06,
+      "loss": 2.016,
+      "step": 7085
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.4709801392427175e-06,
+      "loss": 1.999,
+      "step": 7090
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.4677488619415896e-06,
+      "loss": 2.0008,
+      "step": 7095
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.464516221513871e-06,
+      "loss": 2.0333,
+      "step": 7100
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.461282219745195e-06,
+      "loss": 2.0019,
+      "step": 7105
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.458046858421949e-06,
+      "loss": 1.9907,
+      "step": 7110
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.45481013933127e-06,
+      "loss": 1.9938,
+      "step": 7115
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.451572064261046e-06,
+      "loss": 2.0281,
+      "step": 7120
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.448332634999914e-06,
+      "loss": 2.0126,
+      "step": 7125
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.44509185333726e-06,
+      "loss": 1.9876,
+      "step": 7130
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.441849721063213e-06,
+      "loss": 1.9826,
+      "step": 7135
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.438606239968652e-06,
+      "loss": 2.0195,
+      "step": 7140
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.435361411845201e-06,
+      "loss": 1.9747,
+      "step": 7145
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.432115238485227e-06,
+      "loss": 2.0263,
+      "step": 7150
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.428867721681837e-06,
+      "loss": 2.0185,
+      "step": 7155
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.425618863228887e-06,
+      "loss": 2.0017,
+      "step": 7160
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.422368664920968e-06,
+      "loss": 2.0155,
+      "step": 7165
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.419117128553415e-06,
+      "loss": 2.0274,
+      "step": 7170
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.415864255922298e-06,
+      "loss": 2.0155,
+      "step": 7175
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.412610048824429e-06,
+      "loss": 1.9825,
+      "step": 7180
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.409354509057355e-06,
+      "loss": 2.0057,
+      "step": 7185
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.4060976384193595e-06,
+      "loss": 2.0173,
+      "step": 7190
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.4028394387094636e-06,
+      "loss": 2.0344,
+      "step": 7195
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.399579911727419e-06,
+      "loss": 1.981,
+      "step": 7200
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.396319059273711e-06,
+      "loss": 1.9931,
+      "step": 7205
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.39305688314956e-06,
+      "loss": 1.9884,
+      "step": 7210
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.389793385156914e-06,
+      "loss": 2.006,
+      "step": 7215
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.3865285670984535e-06,
+      "loss": 2.0131,
+      "step": 7220
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.383262430777586e-06,
+      "loss": 2.0033,
+      "step": 7225
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.3799949779984515e-06,
+      "loss": 1.9844,
+      "step": 7230
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.376726210565912e-06,
+      "loss": 2.0183,
+      "step": 7235
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.373456130285559e-06,
+      "loss": 2.0306,
+      "step": 7240
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.370184738963707e-06,
+      "loss": 2.0057,
+      "step": 7245
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.366912038407397e-06,
+      "loss": 2.0269,
+      "step": 7250
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.363638030424392e-06,
+      "loss": 2.0185,
+      "step": 7255
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.360362716823175e-06,
+      "loss": 1.9855,
+      "step": 7260
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.3570860994129556e-06,
+      "loss": 2.0448,
+      "step": 7265
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.353808180003658e-06,
+      "loss": 2.0077,
+      "step": 7270
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.35052896040593e-06,
+      "loss": 2.0157,
+      "step": 7275
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.347248442431134e-06,
+      "loss": 1.9956,
+      "step": 7280
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.343966627891353e-06,
+      "loss": 2.0052,
+      "step": 7285
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.340683518599382e-06,
+      "loss": 1.9903,
+      "step": 7290
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.337399116368734e-06,
+      "loss": 2.0329,
+      "step": 7295
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.334113423013638e-06,
+      "loss": 2.0009,
+      "step": 7300
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.330826440349032e-06,
+      "loss": 2.0359,
+      "step": 7305
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.327538170190569e-06,
+      "loss": 2.0181,
+      "step": 7310
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.324248614354612e-06,
+      "loss": 1.9776,
+      "step": 7315
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.320957774658233e-06,
+      "loss": 1.9886,
+      "step": 7320
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.317665652919217e-06,
+      "loss": 2.0081,
+      "step": 7325
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.3143722509560534e-06,
+      "loss": 2.0041,
+      "step": 7330
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.311077570587941e-06,
+      "loss": 2.0067,
+      "step": 7335
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.307781613634783e-06,
+      "loss": 2.0223,
+      "step": 7340
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.30448438191719e-06,
+      "loss": 2.0031,
+      "step": 7345
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.301185877256476e-06,
+      "loss": 2.0019,
+      "step": 7350
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.2978861014746564e-06,
+      "loss": 1.9874,
+      "step": 7355
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.294585056394448e-06,
+      "loss": 2.0098,
+      "step": 7360
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.291282743839272e-06,
+      "loss": 2.0089,
+      "step": 7365
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.287979165633251e-06,
+      "loss": 2.0001,
+      "step": 7370
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.284674323601202e-06,
+      "loss": 2.0162,
+      "step": 7375
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.281368219568643e-06,
+      "loss": 1.9982,
+      "step": 7380
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.278060855361786e-06,
+      "loss": 1.9988,
+      "step": 7385
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.274752232807544e-06,
+      "loss": 1.9903,
+      "step": 7390
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.271442353733521e-06,
+      "loss": 2.038,
+      "step": 7395
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.268131219968018e-06,
+      "loss": 2.0099,
+      "step": 7400
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.264818833340026e-06,
+      "loss": 1.9982,
+      "step": 7405
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.261505195679232e-06,
+      "loss": 1.9957,
+      "step": 7410
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.258190308816009e-06,
+      "loss": 2.0257,
+      "step": 7415
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.254874174581424e-06,
+      "loss": 1.9842,
+      "step": 7420
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.251556794807233e-06,
+      "loss": 2.017,
+      "step": 7425
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.2482381713258764e-06,
+      "loss": 1.9822,
+      "step": 7430
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.244918305970484e-06,
+      "loss": 2.0103,
+      "step": 7435
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.241597200574875e-06,
+      "loss": 1.9921,
+      "step": 7440
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.238274856973546e-06,
+      "loss": 2.0075,
+      "step": 7445
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.234951277001684e-06,
+      "loss": 2.008,
+      "step": 7450
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.231626462495156e-06,
+      "loss": 2.0072,
+      "step": 7455
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.228300415290512e-06,
+      "loss": 1.9986,
+      "step": 7460
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.224973137224982e-06,
+      "loss": 1.9989,
+      "step": 7465
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.221644630136477e-06,
+      "loss": 1.9788,
+      "step": 7470
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.218314895863585e-06,
+      "loss": 1.9995,
+      "step": 7475
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.214983936245576e-06,
+      "loss": 1.9678,
+      "step": 7480
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.211651753122389e-06,
+      "loss": 2.0028,
+      "step": 7485
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.2083183483346485e-06,
+      "loss": 2.018,
+      "step": 7490
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.2049837237236465e-06,
+      "loss": 1.9876,
+      "step": 7495
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.20164788113135e-06,
+      "loss": 1.9934,
+      "step": 7500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.198310822400406e-06,
+      "loss": 1.9921,
+      "step": 7505
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.194972549374122e-06,
+      "loss": 1.9901,
+      "step": 7510
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.191633063896484e-06,
+      "loss": 1.9845,
+      "step": 7515
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.188292367812144e-06,
+      "loss": 1.9931,
+      "step": 7520
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.184950462966427e-06,
+      "loss": 1.9989,
+      "step": 7525
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.18160735120532e-06,
+      "loss": 2.0097,
+      "step": 7530
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.17826303437548e-06,
+      "loss": 1.993,
+      "step": 7535
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.174917514324231e-06,
+      "loss": 2.013,
+      "step": 7540
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.171570792899558e-06,
+      "loss": 2.0191,
+      "step": 7545
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.1682228719501125e-06,
+      "loss": 1.9734,
+      "step": 7550
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.164873753325205e-06,
+      "loss": 2.016,
+      "step": 7555
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.161523438874814e-06,
+      "loss": 2.0192,
+      "step": 7560
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.158171930449572e-06,
+      "loss": 1.9988,
+      "step": 7565
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.1548192299007745e-06,
+      "loss": 1.9943,
+      "step": 7570
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.151465339080375e-06,
+      "loss": 1.9643,
+      "step": 7575
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.148110259840982e-06,
+      "loss": 2.0152,
+      "step": 7580
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.1447539940358654e-06,
+      "loss": 1.9862,
+      "step": 7585
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.141396543518947e-06,
+      "loss": 1.9872,
+      "step": 7590
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.138037910144803e-06,
+      "loss": 1.9787,
+      "step": 7595
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.134678095768663e-06,
+      "loss": 1.9902,
+      "step": 7600
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.13131710224641e-06,
+      "loss": 2.0028,
+      "step": 7605
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.12795493143458e-06,
+      "loss": 2.0089,
+      "step": 7610
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.124591585190356e-06,
+      "loss": 1.9795,
+      "step": 7615
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.121227065371571e-06,
+      "loss": 2.0059,
+      "step": 7620
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.117861373836707e-06,
+      "loss": 1.9922,
+      "step": 7625
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.114494512444893e-06,
+      "loss": 1.9897,
+      "step": 7630
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.111126483055906e-06,
+      "loss": 1.9914,
+      "step": 7635
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.107757287530163e-06,
+      "loss": 1.9838,
+      "step": 7640
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.104386927728731e-06,
+      "loss": 2.0133,
+      "step": 7645
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.101015405513318e-06,
+      "loss": 2.0182,
+      "step": 7650
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.097642722746274e-06,
+      "loss": 2.0091,
+      "step": 7655
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.094268881290589e-06,
+      "loss": 1.9632,
+      "step": 7660
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.090893883009893e-06,
+      "loss": 1.9789,
+      "step": 7665
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.087517729768457e-06,
+      "loss": 2.0037,
+      "step": 7670
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.0841404234311875e-06,
+      "loss": 1.9785,
+      "step": 7675
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.0807619658636314e-06,
+      "loss": 1.9582,
+      "step": 7680
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.077382358931969e-06,
+      "loss": 2.0369,
+      "step": 7685
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.0740016045030154e-06,
+      "loss": 2.0072,
+      "step": 7690
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.070619704444218e-06,
+      "loss": 1.9724,
+      "step": 7695
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.067236660623663e-06,
+      "loss": 1.9987,
+      "step": 7700
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.063852474910062e-06,
+      "loss": 1.9887,
+      "step": 7705
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.0604671491727606e-06,
+      "loss": 1.9973,
+      "step": 7710
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.057080685281732e-06,
+      "loss": 1.9913,
+      "step": 7715
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.053693085107583e-06,
+      "loss": 1.996,
+      "step": 7720
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.050304350521539e-06,
+      "loss": 1.9841,
+      "step": 7725
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.046914483395463e-06,
+      "loss": 1.9849,
+      "step": 7730
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.043523485601834e-06,
+      "loss": 2.0229,
+      "step": 7735
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.040131359013762e-06,
+      "loss": 2.02,
+      "step": 7740
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.036738105504977e-06,
+      "loss": 2.0106,
+      "step": 7745
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.033343726949833e-06,
+      "loss": 2.0171,
+      "step": 7750
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.029948225223307e-06,
+      "loss": 1.9884,
+      "step": 7755
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.026551602200989e-06,
+      "loss": 1.9771,
+      "step": 7760
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.0231538597590995e-06,
+      "loss": 1.9846,
+      "step": 7765
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.019754999774471e-06,
+      "loss": 2.0004,
+      "step": 7770
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.01635502412455e-06,
+      "loss": 1.9965,
+      "step": 7775
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.012953934687408e-06,
+      "loss": 1.9818,
+      "step": 7780
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.009551733341726e-06,
+      "loss": 1.9854,
+      "step": 7785
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.006148421966798e-06,
+      "loss": 1.9891,
+      "step": 7790
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.002744002442537e-06,
+      "loss": 1.9974,
+      "step": 7795
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.99933847664946e-06,
+      "loss": 1.9736,
+      "step": 7800
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.995931846468703e-06,
+      "loss": 2.0062,
+      "step": 7805
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.992524113782006e-06,
+      "loss": 2.0091,
+      "step": 7810
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.989115280471723e-06,
+      "loss": 2.0161,
+      "step": 7815
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.985705348420813e-06,
+      "loss": 1.9886,
+      "step": 7820
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.9822943195128415e-06,
+      "loss": 2.0108,
+      "step": 7825
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.9788821956319816e-06,
+      "loss": 1.9982,
+      "step": 7830
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.975468978663011e-06,
+      "loss": 1.9631,
+      "step": 7835
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.972054670491309e-06,
+      "loss": 2.0192,
+      "step": 7840
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.9686392730028606e-06,
+      "loss": 1.9989,
+      "step": 7845
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.965222788084252e-06,
+      "loss": 1.9527,
+      "step": 7850
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.961805217622668e-06,
+      "loss": 1.9641,
+      "step": 7855
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.958386563505895e-06,
+      "loss": 1.9974,
+      "step": 7860
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.954966827622318e-06,
+      "loss": 2.0022,
+      "step": 7865
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.951546011860917e-06,
+      "loss": 2.0097,
+      "step": 7870
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.948124118111271e-06,
+      "loss": 1.9886,
+      "step": 7875
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.944701148263555e-06,
+      "loss": 1.9993,
+      "step": 7880
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.941277104208536e-06,
+      "loss": 1.9831,
+      "step": 7885
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.937851987837577e-06,
+      "loss": 2.0075,
+      "step": 7890
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.93442580104263e-06,
+      "loss": 1.9956,
+      "step": 7895
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.930998545716243e-06,
+      "loss": 1.9935,
+      "step": 7900
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.92757022375155e-06,
+      "loss": 2.0097,
+      "step": 7905
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.924140837042275e-06,
+      "loss": 1.9595,
+      "step": 7910
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.920710387482729e-06,
+      "loss": 2.0003,
+      "step": 7915
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.917278876967817e-06,
+      "loss": 1.9973,
+      "step": 7920
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.913846307393021e-06,
+      "loss": 1.975,
+      "step": 7925
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.910412680654413e-06,
+      "loss": 2.0013,
+      "step": 7930
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.906977998648648e-06,
+      "loss": 2.0093,
+      "step": 7935
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.903542263272962e-06,
+      "loss": 1.9934,
+      "step": 7940
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.9001054764251785e-06,
+      "loss": 1.9908,
+      "step": 7945
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.896667640003694e-06,
+      "loss": 1.9771,
+      "step": 7950
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.8932287559074905e-06,
+      "loss": 2.0207,
+      "step": 7955
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.889788826036129e-06,
+      "loss": 1.9817,
+      "step": 7960
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.886347852289743e-06,
+      "loss": 1.9827,
+      "step": 7965
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.882905836569046e-06,
+      "loss": 1.9789,
+      "step": 7970
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.879462780775327e-06,
+      "loss": 1.9272,
+      "step": 7975
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.876018686810451e-06,
+      "loss": 1.9982,
+      "step": 7980
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.872573556576853e-06,
+      "loss": 1.9807,
+      "step": 7985
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.869127391977541e-06,
+      "loss": 1.9922,
+      "step": 7990
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.865680194916099e-06,
+      "loss": 1.9792,
+      "step": 7995
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.8622319672966756e-06,
+      "loss": 1.9966,
+      "step": 8000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.858782711023989e-06,
+      "loss": 2.0134,
+      "step": 8005
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.855332428003332e-06,
+      "loss": 1.9924,
+      "step": 8010
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.8518811201405555e-06,
+      "loss": 2.0311,
+      "step": 8015
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.848428789342084e-06,
+      "loss": 1.979,
+      "step": 8020
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.844975437514904e-06,
+      "loss": 1.9724,
+      "step": 8025
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.841521066566563e-06,
+      "loss": 2.0191,
+      "step": 8030
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.838065678405179e-06,
+      "loss": 1.9631,
+      "step": 8035
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.834609274939424e-06,
+      "loss": 1.9961,
+      "step": 8040
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.831151858078536e-06,
+      "loss": 1.9915,
+      "step": 8045
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.82769342973231e-06,
+      "loss": 1.9823,
+      "step": 8050
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.824233991811099e-06,
+      "loss": 2.01,
+      "step": 8055
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.820773546225819e-06,
+      "loss": 1.9905,
+      "step": 8060
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.817312094887938e-06,
+      "loss": 1.9954,
+      "step": 8065
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.813849639709477e-06,
+      "loss": 1.9964,
+      "step": 8070
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.810386182603021e-06,
+      "loss": 1.978,
+      "step": 8075
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.806921725481697e-06,
+      "loss": 2.0079,
+      "step": 8080
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.803456270259195e-06,
+      "loss": 2.0086,
+      "step": 8085
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.7999898188497445e-06,
+      "loss": 1.9693,
+      "step": 8090
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.796522373168138e-06,
+      "loss": 1.9886,
+      "step": 8095
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.79305393512971e-06,
+      "loss": 1.9959,
+      "step": 8100
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.789584506650342e-06,
+      "loss": 1.9775,
+      "step": 8105
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.786114089646467e-06,
+      "loss": 1.9665,
+      "step": 8110
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.782642686035061e-06,
+      "loss": 1.9786,
+      "step": 8115
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.779170297733645e-06,
+      "loss": 1.9692,
+      "step": 8120
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.775696926660287e-06,
+      "loss": 1.9933,
+      "step": 8125
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.772222574733593e-06,
+      "loss": 2.0001,
+      "step": 8130
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.768747243872716e-06,
+      "loss": 1.9945,
+      "step": 8135
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.765270935997348e-06,
+      "loss": 1.9988,
+      "step": 8140
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.761793653027717e-06,
+      "loss": 1.9617,
+      "step": 8145
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.758315396884595e-06,
+      "loss": 2.0309,
+      "step": 8150
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.754836169489287e-06,
+      "loss": 1.9955,
+      "step": 8155
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.751355972763638e-06,
+      "loss": 1.9837,
+      "step": 8160
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.747874808630027e-06,
+      "loss": 1.9931,
+      "step": 8165
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.744392679011367e-06,
+      "loss": 2.0022,
+      "step": 8170
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.740909585831106e-06,
+      "loss": 2.0008,
+      "step": 8175
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.73742553101322e-06,
+      "loss": 2.0063,
+      "step": 8180
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.73394051648222e-06,
+      "loss": 2.0157,
+      "step": 8185
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.73045454416315e-06,
+      "loss": 1.9619,
+      "step": 8190
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.726967615981574e-06,
+      "loss": 1.9828,
+      "step": 8195
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.7234797338635934e-06,
+      "loss": 1.9859,
+      "step": 8200
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.719990899735832e-06,
+      "loss": 1.983,
+      "step": 8205
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.716501115525439e-06,
+      "loss": 2.0125,
+      "step": 8210
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.713010383160089e-06,
+      "loss": 2.029,
+      "step": 8215
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.709518704567981e-06,
+      "loss": 2.0059,
+      "step": 8220
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.706026081677835e-06,
+      "loss": 2.0031,
+      "step": 8225
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.702532516418896e-06,
+      "loss": 1.9851,
+      "step": 8230
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.699038010720928e-06,
+      "loss": 2.012,
+      "step": 8235
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.695542566514212e-06,
+      "loss": 1.9834,
+      "step": 8240
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.69204618572955e-06,
+      "loss": 2.0009,
+      "step": 8245
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.6885488702982615e-06,
+      "loss": 1.9981,
+      "step": 8250
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.685050622152181e-06,
+      "loss": 2.0043,
+      "step": 8255
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.681551443223657e-06,
+      "loss": 2.0063,
+      "step": 8260
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.678051335445558e-06,
+      "loss": 1.9913,
+      "step": 8265
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.674550300751258e-06,
+      "loss": 2.0045,
+      "step": 8270
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.671048341074648e-06,
+      "loss": 2.0155,
+      "step": 8275
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.6675454583501275e-06,
+      "loss": 1.9791,
+      "step": 8280
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.664041654512607e-06,
+      "loss": 2.0007,
+      "step": 8285
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.660536931497507e-06,
+      "loss": 2.003,
+      "step": 8290
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.6570312912407506e-06,
+      "loss": 1.9856,
+      "step": 8295
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.653524735678776e-06,
+      "loss": 1.9841,
+      "step": 8300
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.650017266748518e-06,
+      "loss": 1.987,
+      "step": 8305
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.646508886387424e-06,
+      "loss": 2.0019,
+      "step": 8310
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.642999596533437e-06,
+      "loss": 2.0045,
+      "step": 8315
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.63948939912501e-06,
+      "loss": 2.0142,
+      "step": 8320
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.635978296101092e-06,
+      "loss": 1.9783,
+      "step": 8325
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.632466289401132e-06,
+      "loss": 1.9905,
+      "step": 8330
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.628953380965085e-06,
+      "loss": 2.0062,
+      "step": 8335
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.625439572733397e-06,
+      "loss": 1.9743,
+      "step": 8340
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.6219248666470115e-06,
+      "loss": 1.9964,
+      "step": 8345
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.618409264647372e-06,
+      "loss": 1.9911,
+      "step": 8350
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.614892768676412e-06,
+      "loss": 1.9903,
+      "step": 8355
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.611375380676565e-06,
+      "loss": 1.9868,
+      "step": 8360
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.607857102590749e-06,
+      "loss": 1.9986,
+      "step": 8365
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.604337936362382e-06,
+      "loss": 2.004,
+      "step": 8370
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.600817883935368e-06,
+      "loss": 1.9736,
+      "step": 8375
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.5972969472541e-06,
+      "loss": 1.9946,
+      "step": 8380
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.593775128263464e-06,
+      "loss": 1.9828,
+      "step": 8385
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.5902524289088274e-06,
+      "loss": 1.9921,
+      "step": 8390
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.5867288511360484e-06,
+      "loss": 1.9791,
+      "step": 8395
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.583204396891466e-06,
+      "loss": 2.0076,
+      "step": 8400
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.579679068121909e-06,
+      "loss": 1.9842,
+      "step": 8405
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.576152866774685e-06,
+      "loss": 2.0052,
+      "step": 8410
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.572625794797587e-06,
+      "loss": 1.9878,
+      "step": 8415
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.569097854138884e-06,
+      "loss": 1.9877,
+      "step": 8420
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.565569046747328e-06,
+      "loss": 2.0154,
+      "step": 8425
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.562039374572149e-06,
+      "loss": 1.9932,
+      "step": 8430
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.558508839563058e-06,
+      "loss": 1.9539,
+      "step": 8435
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.554977443670236e-06,
+      "loss": 1.9725,
+      "step": 8440
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.5514451888443475e-06,
+      "loss": 1.9767,
+      "step": 8445
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.547912077036524e-06,
+      "loss": 1.9583,
+      "step": 8450
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.544378110198375e-06,
+      "loss": 1.9773,
+      "step": 8455
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.540843290281981e-06,
+      "loss": 1.9652,
+      "step": 8460
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.537307619239891e-06,
+      "loss": 1.9713,
+      "step": 8465
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.53377109902513e-06,
+      "loss": 1.9933,
+      "step": 8470
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.530233731591188e-06,
+      "loss": 1.9643,
+      "step": 8475
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.526695518892024e-06,
+      "loss": 2.001,
+      "step": 8480
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.523156462882062e-06,
+      "loss": 1.9766,
+      "step": 8485
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.519616565516195e-06,
+      "loss": 1.9835,
+      "step": 8490
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.516075828749777e-06,
+      "loss": 2.002,
+      "step": 8495
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.512534254538632e-06,
+      "loss": 1.9994,
+      "step": 8500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.50899184483904e-06,
+      "loss": 1.981,
+      "step": 8505
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.505448601607747e-06,
+      "loss": 2.0205,
+      "step": 8510
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.501904526801954e-06,
+      "loss": 1.9982,
+      "step": 8515
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.498359622379327e-06,
+      "loss": 1.9891,
+      "step": 8520
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.494813890297988e-06,
+      "loss": 2.0008,
+      "step": 8525
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.491267332516518e-06,
+      "loss": 1.9897,
+      "step": 8530
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.487719950993947e-06,
+      "loss": 1.9725,
+      "step": 8535
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.48417174768977e-06,
+      "loss": 2.0037,
+      "step": 8540
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.480622724563931e-06,
+      "loss": 2.0047,
+      "step": 8545
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.4770728835768246e-06,
+      "loss": 1.9787,
+      "step": 8550
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.473522226689303e-06,
+      "loss": 1.9967,
+      "step": 8555
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.469970755862661e-06,
+      "loss": 1.99,
+      "step": 8560
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.466418473058653e-06,
+      "loss": 1.9904,
+      "step": 8565
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.462865380239474e-06,
+      "loss": 1.9788,
+      "step": 8570
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.459311479367771e-06,
+      "loss": 1.9752,
+      "step": 8575
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.455756772406631e-06,
+      "loss": 1.9982,
+      "step": 8580
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.4522012613195974e-06,
+      "loss": 1.9844,
+      "step": 8585
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.4486449480706455e-06,
+      "loss": 1.985,
+      "step": 8590
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.445087834624203e-06,
+      "loss": 1.9753,
+      "step": 8595
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.441529922945134e-06,
+      "loss": 1.9622,
+      "step": 8600
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.437971214998744e-06,
+      "loss": 1.9593,
+      "step": 8605
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.434411712750784e-06,
+      "loss": 1.988,
+      "step": 8610
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.430851418167435e-06,
+      "loss": 1.969,
+      "step": 8615
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.427290333215323e-06,
+      "loss": 2.0035,
+      "step": 8620
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.423728459861508e-06,
+      "loss": 1.9808,
+      "step": 8625
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.420165800073483e-06,
+      "loss": 1.9875,
+      "step": 8630
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.4166023558191815e-06,
+      "loss": 2.0004,
+      "step": 8635
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.413038129066963e-06,
+      "loss": 1.9462,
+      "step": 8640
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.409473121785624e-06,
+      "loss": 1.9693,
+      "step": 8645
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.405907335944391e-06,
+      "loss": 1.976,
+      "step": 8650
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.402340773512921e-06,
+      "loss": 2.002,
+      "step": 8655
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.398773436461299e-06,
+      "loss": 1.9716,
+      "step": 8660
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.395205326760037e-06,
+      "loss": 1.9865,
+      "step": 8665
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.3916364463800774e-06,
+      "loss": 1.991,
+      "step": 8670
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.388066797292783e-06,
+      "loss": 1.9831,
+      "step": 8675
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.384496381469948e-06,
+      "loss": 1.9809,
+      "step": 8680
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.380925200883783e-06,
+      "loss": 1.9967,
+      "step": 8685
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.377353257506927e-06,
+      "loss": 1.9727,
+      "step": 8690
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.3737805533124365e-06,
+      "loss": 1.9953,
+      "step": 8695
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.370207090273789e-06,
+      "loss": 1.9924,
+      "step": 8700
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.3666328703648825e-06,
+      "loss": 2.0036,
+      "step": 8705
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.363057895560032e-06,
+      "loss": 1.9921,
+      "step": 8710
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.35948216783397e-06,
+      "loss": 2.005,
+      "step": 8715
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.3559056891618445e-06,
+      "loss": 2.0041,
+      "step": 8720
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.352328461519217e-06,
+      "loss": 1.9923,
+      "step": 8725
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.348750486882067e-06,
+      "loss": 1.959,
+      "step": 8730
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.3451717672267825e-06,
+      "loss": 1.992,
+      "step": 8735
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.341592304530162e-06,
+      "loss": 1.998,
+      "step": 8740
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.3380121007694206e-06,
+      "loss": 1.9945,
+      "step": 8745
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.334431157922176e-06,
+      "loss": 1.9784,
+      "step": 8750
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.330849477966459e-06,
+      "loss": 1.9659,
+      "step": 8755
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.327267062880704e-06,
+      "loss": 1.9571,
+      "step": 8760
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.323683914643754e-06,
+      "loss": 1.9696,
+      "step": 8765
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.320100035234856e-06,
+      "loss": 1.9765,
+      "step": 8770
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.3165154266336605e-06,
+      "loss": 1.956,
+      "step": 8775
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.312930090820222e-06,
+      "loss": 1.9935,
+      "step": 8780
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.309344029774994e-06,
+      "loss": 1.9981,
+      "step": 8785
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.305757245478835e-06,
+      "loss": 1.985,
+      "step": 8790
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.302169739913e-06,
+      "loss": 2.0212,
+      "step": 8795
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.298581515059141e-06,
+      "loss": 1.9576,
+      "step": 8800
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.29499257289931e-06,
+      "loss": 1.9817,
+      "step": 8805
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.291402915415958e-06,
+      "loss": 1.9487,
+      "step": 8810
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.28781254459192e-06,
+      "loss": 2.0155,
+      "step": 8815
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.284221462410442e-06,
+      "loss": 1.9884,
+      "step": 8820
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.2806296708551466e-06,
+      "loss": 1.953,
+      "step": 8825
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.277037171910058e-06,
+      "loss": 1.9702,
+      "step": 8830
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.273443967559587e-06,
+      "loss": 1.9649,
+      "step": 8835
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.269850059788538e-06,
+      "loss": 2.0091,
+      "step": 8840
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.2662554505820976e-06,
+      "loss": 1.9947,
+      "step": 8845
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.262660141925846e-06,
+      "loss": 1.9836,
+      "step": 8850
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.259064135805748e-06,
+      "loss": 1.9687,
+      "step": 8855
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.2554674342081515e-06,
+      "loss": 1.9822,
+      "step": 8860
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.2518700391197904e-06,
+      "loss": 1.9821,
+      "step": 8865
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.248271952527781e-06,
+      "loss": 1.9773,
+      "step": 8870
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.244673176419624e-06,
+      "loss": 1.9922,
+      "step": 8875
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.241073712783195e-06,
+      "loss": 2.0008,
+      "step": 8880
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.237473563606758e-06,
+      "loss": 1.9733,
+      "step": 8885
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.23387273087895e-06,
+      "loss": 1.9615,
+      "step": 8890
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.230271216588783e-06,
+      "loss": 1.9717,
+      "step": 8895
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.226669022725651e-06,
+      "loss": 1.9901,
+      "step": 8900
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.223066151279323e-06,
+      "loss": 1.9526,
+      "step": 8905
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.219462604239939e-06,
+      "loss": 1.9719,
+      "step": 8910
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.215858383598013e-06,
+      "loss": 1.9942,
+      "step": 8915
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.212253491344435e-06,
+      "loss": 1.962,
+      "step": 8920
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.208647929470461e-06,
+      "loss": 1.9649,
+      "step": 8925
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.205041699967716e-06,
+      "loss": 1.9701,
+      "step": 8930
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.201434804828203e-06,
+      "loss": 1.9817,
+      "step": 8935
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.197827246044281e-06,
+      "loss": 1.9765,
+      "step": 8940
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.194219025608683e-06,
+      "loss": 1.9619,
+      "step": 8945
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.190610145514505e-06,
+      "loss": 1.9898,
+      "step": 8950
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.187000607755208e-06,
+      "loss": 1.9922,
+      "step": 8955
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.183390414324616e-06,
+      "loss": 1.9367,
+      "step": 8960
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.179779567216915e-06,
+      "loss": 1.9784,
+      "step": 8965
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.176168068426651e-06,
+      "loss": 2.0056,
+      "step": 8970
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.172555919948733e-06,
+      "loss": 1.9715,
+      "step": 8975
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.168943123778427e-06,
+      "loss": 1.9771,
+      "step": 8980
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.1653296819113554e-06,
+      "loss": 1.968,
+      "step": 8985
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.161715596343499e-06,
+      "loss": 1.9768,
+      "step": 8990
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.158100869071195e-06,
+      "loss": 1.994,
+      "step": 8995
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.154485502091134e-06,
+      "loss": 1.9635,
+      "step": 9000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.15086949740036e-06,
+      "loss": 1.9527,
+      "step": 9005
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.147252856996269e-06,
+      "loss": 1.9764,
+      "step": 9010
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.143635582876607e-06,
+      "loss": 1.9459,
+      "step": 9015
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.140017677039472e-06,
+      "loss": 1.9865,
+      "step": 9020
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.1363991414833115e-06,
+      "loss": 1.9568,
+      "step": 9025
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.132779978206917e-06,
+      "loss": 1.9682,
+      "step": 9030
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.129160189209434e-06,
+      "loss": 1.9724,
+      "step": 9035
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.125539776490346e-06,
+      "loss": 1.9855,
+      "step": 9040
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.121918742049483e-06,
+      "loss": 1.9933,
+      "step": 9045
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.118297087887019e-06,
+      "loss": 1.9742,
+      "step": 9050
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.1146748160034764e-06,
+      "loss": 2.0002,
+      "step": 9055
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.111051928399707e-06,
+      "loss": 1.9731,
+      "step": 9060
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.107428427076912e-06,
+      "loss": 1.9893,
+      "step": 9065
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.10380431403663e-06,
+      "loss": 2.0083,
+      "step": 9070
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.100179591280733e-06,
+      "loss": 1.9849,
+      "step": 9075
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.096554260811433e-06,
+      "loss": 2.0103,
+      "step": 9080
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.092928324631278e-06,
+      "loss": 1.9736,
+      "step": 9085
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.089301784743151e-06,
+      "loss": 1.9859,
+      "step": 9090
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.0856746431502675e-06,
+      "loss": 2.0031,
+      "step": 9095
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.082046901856176e-06,
+      "loss": 1.9699,
+      "step": 9100
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.078418562864753e-06,
+      "loss": 1.974,
+      "step": 9105
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.074789628180211e-06,
+      "loss": 1.9773,
+      "step": 9110
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.071160099807085e-06,
+      "loss": 1.9566,
+      "step": 9115
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.067529979750244e-06,
+      "loss": 1.9909,
+      "step": 9120
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.0638992700148804e-06,
+      "loss": 1.9829,
+      "step": 9125
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.060267972606514e-06,
+      "loss": 2.0108,
+      "step": 9130
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.056636089530987e-06,
+      "loss": 1.9997,
+      "step": 9135
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.053003622794468e-06,
+      "loss": 1.9722,
+      "step": 9140
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.049370574403445e-06,
+      "loss": 1.9877,
+      "step": 9145
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.0457369463647285e-06,
+      "loss": 1.9734,
+      "step": 9150
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.042102740685449e-06,
+      "loss": 1.9751,
+      "step": 9155
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.03846795937306e-06,
+      "loss": 1.9854,
+      "step": 9160
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.034832604435325e-06,
+      "loss": 1.9872,
+      "step": 9165
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.031196677880332e-06,
+      "loss": 1.9753,
+      "step": 9170
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.02756018171648e-06,
+      "loss": 1.979,
+      "step": 9175
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.0239231179524826e-06,
+      "loss": 1.9935,
+      "step": 9180
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.020285488597373e-06,
+      "loss": 1.9911,
+      "step": 9185
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.016647295660487e-06,
+      "loss": 1.9907,
+      "step": 9190
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.013008541151482e-06,
+      "loss": 1.9941,
+      "step": 9195
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.009369227080317e-06,
+      "loss": 1.9868,
+      "step": 9200
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.005729355457266e-06,
+      "loss": 2.0119,
+      "step": 9205
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.002088928292908e-06,
+      "loss": 1.9646,
+      "step": 9210
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.998447947598129e-06,
+      "loss": 1.9715,
+      "step": 9215
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.994806415384123e-06,
+      "loss": 1.9947,
+      "step": 9220
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.991164333662384e-06,
+      "loss": 2.0062,
+      "step": 9225
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.987521704444715e-06,
+      "loss": 1.9945,
+      "step": 9230
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.983878529743219e-06,
+      "loss": 1.9727,
+      "step": 9235
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.980234811570298e-06,
+      "loss": 1.9873,
+      "step": 9240
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.976590551938658e-06,
+      "loss": 1.9874,
+      "step": 9245
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.972945752861303e-06,
+      "loss": 1.9928,
+      "step": 9250
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.969300416351535e-06,
+      "loss": 1.9395,
+      "step": 9255
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.9656545444229496e-06,
+      "loss": 1.956,
+      "step": 9260
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.962008139089445e-06,
+      "loss": 1.9718,
+      "step": 9265
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.958361202365206e-06,
+      "loss": 1.9776,
+      "step": 9270
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.954713736264717e-06,
+      "loss": 1.9659,
+      "step": 9275
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.951065742802753e-06,
+      "loss": 1.9851,
+      "step": 9280
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.947417223994378e-06,
+      "loss": 1.9666,
+      "step": 9285
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.943768181854952e-06,
+      "loss": 2.0164,
+      "step": 9290
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.940118618400114e-06,
+      "loss": 1.9672,
+      "step": 9295
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.936468535645805e-06,
+      "loss": 1.9869,
+      "step": 9300
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.932817935608238e-06,
+      "loss": 1.969,
+      "step": 9305
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.929166820303924e-06,
+      "loss": 1.9826,
+      "step": 9310
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.925515191749652e-06,
+      "loss": 1.9664,
+      "step": 9315
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.921863051962495e-06,
+      "loss": 1.9918,
+      "step": 9320
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.91821040295981e-06,
+      "loss": 1.9631,
+      "step": 9325
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.914557246759234e-06,
+      "loss": 1.9729,
+      "step": 9330
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.910903585378686e-06,
+      "loss": 1.9866,
+      "step": 9335
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.907249420836362e-06,
+      "loss": 1.9825,
+      "step": 9340
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.9035947551507365e-06,
+      "loss": 1.9702,
+      "step": 9345
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.899939590340562e-06,
+      "loss": 1.9639,
+      "step": 9350
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.896283928424865e-06,
+      "loss": 1.978,
+      "step": 9355
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.8926277714229474e-06,
+      "loss": 1.9793,
+      "step": 9360
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.8889711213543855e-06,
+      "loss": 1.9723,
+      "step": 9365
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.885313980239026e-06,
+      "loss": 1.9744,
+      "step": 9370
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.881656350096988e-06,
+      "loss": 2.0029,
+      "step": 9375
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.877998232948663e-06,
+      "loss": 1.9846,
+      "step": 9380
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.874339630814705e-06,
+      "loss": 1.9599,
+      "step": 9385
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.870680545716043e-06,
+      "loss": 1.971,
+      "step": 9390
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.86702097967387e-06,
+      "loss": 1.9707,
+      "step": 9395
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.863360934709641e-06,
+      "loss": 1.9743,
+      "step": 9400
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.859700412845082e-06,
+      "loss": 1.9717,
+      "step": 9405
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8560394161021795e-06,
+      "loss": 1.9551,
+      "step": 9410
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.852377946503182e-06,
+      "loss": 1.9701,
+      "step": 9415
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.848716006070598e-06,
+      "loss": 2.0204,
+      "step": 9420
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.845053596827197e-06,
+      "loss": 1.994,
+      "step": 9425
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.841390720796012e-06,
+      "loss": 1.9405,
+      "step": 9430
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.837727380000324e-06,
+      "loss": 1.9919,
+      "step": 9435
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8340635764636825e-06,
+      "loss": 1.9938,
+      "step": 9440
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.830399312209882e-06,
+      "loss": 1.9886,
+      "step": 9445
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8267345892629805e-06,
+      "loss": 1.9728,
+      "step": 9450
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.82306940964728e-06,
+      "loss": 1.942,
+      "step": 9455
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.819403775387344e-06,
+      "loss": 1.9842,
+      "step": 9460
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.81573768850798e-06,
+      "loss": 1.9626,
+      "step": 9465
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.81207115103425e-06,
+      "loss": 1.9834,
+      "step": 9470
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.808404164991463e-06,
+      "loss": 1.996,
+      "step": 9475
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.804736732405176e-06,
+      "loss": 1.9791,
+      "step": 9480
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.801068855301191e-06,
+      "loss": 1.9976,
+      "step": 9485
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.79740053570556e-06,
+      "loss": 1.9961,
+      "step": 9490
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.793731775644577e-06,
+      "loss": 1.9619,
+      "step": 9495
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.790062577144775e-06,
+      "loss": 1.9783,
+      "step": 9500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.786392942232938e-06,
+      "loss": 1.9539,
+      "step": 9505
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.782722872936084e-06,
+      "loss": 1.9873,
+      "step": 9510
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.779052371281473e-06,
+      "loss": 1.9561,
+      "step": 9515
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.775381439296605e-06,
+      "loss": 1.9743,
+      "step": 9520
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.771710079009215e-06,
+      "loss": 1.954,
+      "step": 9525
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.7680382924472775e-06,
+      "loss": 1.9905,
+      "step": 9530
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.764366081639e-06,
+      "loss": 1.9617,
+      "step": 9535
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.760693448612829e-06,
+      "loss": 1.9725,
+      "step": 9540
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.757020395397438e-06,
+      "loss": 1.9379,
+      "step": 9545
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.753346924021734e-06,
+      "loss": 1.9801,
+      "step": 9550
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.7496730365148616e-06,
+      "loss": 1.9778,
+      "step": 9555
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.745998734906185e-06,
+      "loss": 1.9681,
+      "step": 9560
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.742324021225305e-06,
+      "loss": 1.97,
+      "step": 9565
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.738648897502046e-06,
+      "loss": 1.9726,
+      "step": 9570
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.734973365766461e-06,
+      "loss": 1.9773,
+      "step": 9575
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.731297428048827e-06,
+      "loss": 1.986,
+      "step": 9580
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.727621086379647e-06,
+      "loss": 1.985,
+      "step": 9585
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.7239443427896436e-06,
+      "loss": 1.9619,
+      "step": 9590
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.720267199309766e-06,
+      "loss": 1.9661,
+      "step": 9595
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.71658965797118e-06,
+      "loss": 1.9686,
+      "step": 9600
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.7129117208052735e-06,
+      "loss": 1.9673,
+      "step": 9605
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.709233389843654e-06,
+      "loss": 1.9764,
+      "step": 9610
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.705554667118144e-06,
+      "loss": 1.9305,
+      "step": 9615
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.701875554660785e-06,
+      "loss": 1.9628,
+      "step": 9620
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.698196054503833e-06,
+      "loss": 1.9746,
+      "step": 9625
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.694516168679755e-06,
+      "loss": 1.981,
+      "step": 9630
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.690835899221235e-06,
+      "loss": 1.9792,
+      "step": 9635
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.687155248161167e-06,
+      "loss": 2.0025,
+      "step": 9640
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6834742175326565e-06,
+      "loss": 1.9769,
+      "step": 9645
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.67979280936902e-06,
+      "loss": 1.9798,
+      "step": 9650
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.676111025703779e-06,
+      "loss": 1.9597,
+      "step": 9655
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.672428868570665e-06,
+      "loss": 1.9939,
+      "step": 9660
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.668746340003617e-06,
+      "loss": 1.9735,
+      "step": 9665
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6650634420367734e-06,
+      "loss": 1.9635,
+      "step": 9670
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6613801767044856e-06,
+      "loss": 1.9868,
+      "step": 9675
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.657696546041298e-06,
+      "loss": 1.9815,
+      "step": 9680
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.654012552081966e-06,
+      "loss": 1.9583,
+      "step": 9685
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6503281968614386e-06,
+      "loss": 1.9536,
+      "step": 9690
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6466434824148685e-06,
+      "loss": 1.9444,
+      "step": 9695
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.642958410777607e-06,
+      "loss": 1.9803,
+      "step": 9700
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.639272983985198e-06,
+      "loss": 1.9531,
+      "step": 9705
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.635587204073386e-06,
+      "loss": 1.9548,
+      "step": 9710
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.631901073078111e-06,
+      "loss": 2.0024,
+      "step": 9715
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.628214593035505e-06,
+      "loss": 1.9615,
+      "step": 9720
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.624527765981891e-06,
+      "loss": 1.9805,
+      "step": 9725
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.620840593953787e-06,
+      "loss": 1.989,
+      "step": 9730
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.617153078987899e-06,
+      "loss": 1.978,
+      "step": 9735
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.613465223121128e-06,
+      "loss": 1.9496,
+      "step": 9740
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.609777028390553e-06,
+      "loss": 1.9599,
+      "step": 9745
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.606088496833452e-06,
+      "loss": 1.9809,
+      "step": 9750
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.60239963048728e-06,
+      "loss": 1.9947,
+      "step": 9755
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.5987104313896814e-06,
+      "loss": 1.9619,
+      "step": 9760
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.595020901578483e-06,
+      "loss": 1.9676,
+      "step": 9765
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.591331043091695e-06,
+      "loss": 1.9997,
+      "step": 9770
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.5876408579675056e-06,
+      "loss": 1.9676,
+      "step": 9775
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.583950348244291e-06,
+      "loss": 1.9542,
+      "step": 9780
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.5802595159606e-06,
+      "loss": 1.9634,
+      "step": 9785
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.576568363155163e-06,
+      "loss": 1.9556,
+      "step": 9790
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.572876891866883e-06,
+      "loss": 1.9675,
+      "step": 9795
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.569185104134845e-06,
+      "loss": 1.9622,
+      "step": 9800
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.565493001998307e-06,
+      "loss": 1.936,
+      "step": 9805
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.561800587496696e-06,
+      "loss": 1.9479,
+      "step": 9810
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.558107862669619e-06,
+      "loss": 1.9681,
+      "step": 9815
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.554414829556848e-06,
+      "loss": 2.0075,
+      "step": 9820
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.550721490198328e-06,
+      "loss": 1.9931,
+      "step": 9825
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.547027846634175e-06,
+      "loss": 1.9539,
+      "step": 9830
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.543333900904669e-06,
+      "loss": 1.9835,
+      "step": 9835
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.539639655050261e-06,
+      "loss": 1.9421,
+      "step": 9840
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.535945111111562e-06,
+      "loss": 1.9307,
+      "step": 9845
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.5322502711293556e-06,
+      "loss": 1.958,
+      "step": 9850
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.5285551371445835e-06,
+      "loss": 1.9659,
+      "step": 9855
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.524859711198349e-06,
+      "loss": 1.9615,
+      "step": 9860
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.5211639953319206e-06,
+      "loss": 1.96,
+      "step": 9865
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.517467991586726e-06,
+      "loss": 1.9589,
+      "step": 9870
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.513771702004349e-06,
+      "loss": 1.9413,
+      "step": 9875
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.5100751286265346e-06,
+      "loss": 1.9734,
+      "step": 9880
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.506378273495181e-06,
+      "loss": 1.9922,
+      "step": 9885
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.502681138652348e-06,
+      "loss": 1.961,
+      "step": 9890
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.498983726140243e-06,
+      "loss": 1.9633,
+      "step": 9895
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.495286038001232e-06,
+      "loss": 1.9643,
+      "step": 9900
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.49158807627783e-06,
+      "loss": 1.9407,
+      "step": 9905
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.487889843012704e-06,
+      "loss": 1.9797,
+      "step": 9910
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.484191340248671e-06,
+      "loss": 1.9686,
+      "step": 9915
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.480492570028699e-06,
+      "loss": 1.9689,
+      "step": 9920
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.476793534395901e-06,
+      "loss": 1.9798,
+      "step": 9925
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.473094235393538e-06,
+      "loss": 1.945,
+      "step": 9930
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.469394675065015e-06,
+      "loss": 1.9695,
+      "step": 9935
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.465694855453884e-06,
+      "loss": 1.9475,
+      "step": 9940
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.461994778603836e-06,
+      "loss": 1.9655,
+      "step": 9945
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.458294446558708e-06,
+      "loss": 1.9462,
+      "step": 9950
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.454593861362478e-06,
+      "loss": 1.9308,
+      "step": 9955
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.45089302505926e-06,
+      "loss": 1.9763,
+      "step": 9960
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.447191939693312e-06,
+      "loss": 1.9575,
+      "step": 9965
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.443490607309025e-06,
+      "loss": 1.9921,
+      "step": 9970
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.439789029950927e-06,
+      "loss": 1.9651,
+      "step": 9975
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.436087209663684e-06,
+      "loss": 1.97,
+      "step": 9980
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.432385148492095e-06,
+      "loss": 1.9567,
+      "step": 9985
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.428682848481091e-06,
+      "loss": 1.9623,
+      "step": 9990
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.424980311675736e-06,
+      "loss": 1.9782,
+      "step": 9995
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.421277540121225e-06,
+      "loss": 1.9521,
+      "step": 10000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.417574535862882e-06,
+      "loss": 1.9699,
+      "step": 10005
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.413871300946158e-06,
+      "loss": 1.9693,
+      "step": 10010
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.4101678374166364e-06,
+      "loss": 2.0016,
+      "step": 10015
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.4064641473200195e-06,
+      "loss": 1.9614,
+      "step": 10020
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.402760232702142e-06,
+      "loss": 1.9428,
+      "step": 10025
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.3990560956089584e-06,
+      "loss": 1.9654,
+      "step": 10030
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.395351738086548e-06,
+      "loss": 1.9809,
+      "step": 10035
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.391647162181109e-06,
+      "loss": 1.9592,
+      "step": 10040
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.387942369938964e-06,
+      "loss": 1.9722,
+      "step": 10045
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.384237363406554e-06,
+      "loss": 1.9608,
+      "step": 10050
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.380532144630435e-06,
+      "loss": 1.9733,
+      "step": 10055
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.376826715657284e-06,
+      "loss": 1.9982,
+      "step": 10060
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.373121078533895e-06,
+      "loss": 1.9704,
+      "step": 10065
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.369415235307174e-06,
+      "loss": 1.9749,
+      "step": 10070
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.36570918802414e-06,
+      "loss": 1.9752,
+      "step": 10075
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.362002938731929e-06,
+      "loss": 1.961,
+      "step": 10080
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.358296489477784e-06,
+      "loss": 1.9718,
+      "step": 10085
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3545898423090626e-06,
+      "loss": 1.9862,
+      "step": 10090
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.350882999273229e-06,
+      "loss": 1.9489,
+      "step": 10095
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.347175962417857e-06,
+      "loss": 1.9899,
+      "step": 10100
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3434687337906245e-06,
+      "loss": 1.9591,
+      "step": 10105
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.33976131543932e-06,
+      "loss": 1.9726,
+      "step": 10110
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.336053709411834e-06,
+      "loss": 1.9621,
+      "step": 10115
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.332345917756161e-06,
+      "loss": 1.9884,
+      "step": 10120
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.328637942520398e-06,
+      "loss": 1.9538,
+      "step": 10125
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.324929785752744e-06,
+      "loss": 1.944,
+      "step": 10130
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.321221449501497e-06,
+      "loss": 1.9882,
+      "step": 10135
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.317512935815055e-06,
+      "loss": 1.9698,
+      "step": 10140
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.313804246741915e-06,
+      "loss": 1.9625,
+      "step": 10145
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.310095384330669e-06,
+      "loss": 1.9594,
+      "step": 10150
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.306386350630005e-06,
+      "loss": 1.9495,
+      "step": 10155
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3026771476887075e-06,
+      "loss": 1.9768,
+      "step": 10160
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.2989677775556535e-06,
+      "loss": 1.9611,
+      "step": 10165
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.29525824227981e-06,
+      "loss": 1.9846,
+      "step": 10170
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.29154854391024e-06,
+      "loss": 1.9413,
+      "step": 10175
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.287838684496092e-06,
+      "loss": 1.9591,
+      "step": 10180
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.284128666086605e-06,
+      "loss": 1.9838,
+      "step": 10185
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.2804184907311066e-06,
+      "loss": 1.966,
+      "step": 10190
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.276708160479013e-06,
+      "loss": 1.961,
+      "step": 10195
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.27299767737982e-06,
+      "loss": 1.9468,
+      "step": 10200
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.269287043483113e-06,
+      "loss": 1.96,
+      "step": 10205
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.265576260838561e-06,
+      "loss": 1.9724,
+      "step": 10210
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.26186533149591e-06,
+      "loss": 1.9651,
+      "step": 10215
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.258154257504991e-06,
+      "loss": 1.9522,
+      "step": 10220
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.254443040915716e-06,
+      "loss": 1.9505,
+      "step": 10225
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.250731683778073e-06,
+      "loss": 1.9781,
+      "step": 10230
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.247020188142128e-06,
+      "loss": 1.9641,
+      "step": 10235
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.2433085560580255e-06,
+      "loss": 1.9572,
+      "step": 10240
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.239596789575982e-06,
+      "loss": 1.934,
+      "step": 10245
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.235884890746293e-06,
+      "loss": 1.9788,
+      "step": 10250
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.232172861619324e-06,
+      "loss": 1.9772,
+      "step": 10255
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.22846070424551e-06,
+      "loss": 1.9764,
+      "step": 10260
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.2247484206753606e-06,
+      "loss": 1.9582,
+      "step": 10265
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.2210360129594565e-06,
+      "loss": 1.9709,
+      "step": 10270
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.217323483148443e-06,
+      "loss": 1.9735,
+      "step": 10275
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.213610833293035e-06,
+      "loss": 1.949,
+      "step": 10280
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.209898065444011e-06,
+      "loss": 1.9482,
+      "step": 10285
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.206185181652219e-06,
+      "loss": 1.9651,
+      "step": 10290
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.202472183968569e-06,
+      "loss": 1.9617,
+      "step": 10295
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.198759074444034e-06,
+      "loss": 1.9502,
+      "step": 10300
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.195045855129648e-06,
+      "loss": 1.9523,
+      "step": 10305
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.191332528076507e-06,
+      "loss": 1.9546,
+      "step": 10310
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.187619095335764e-06,
+      "loss": 1.9699,
+      "step": 10315
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.183905558958634e-06,
+      "loss": 1.9681,
+      "step": 10320
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.180191920996387e-06,
+      "loss": 1.9484,
+      "step": 10325
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.176478183500346e-06,
+      "loss": 1.9716,
+      "step": 10330
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.172764348521899e-06,
+      "loss": 1.9528,
+      "step": 10335
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.169050418112476e-06,
+      "loss": 1.9567,
+      "step": 10340
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.165336394323569e-06,
+      "loss": 1.9597,
+      "step": 10345
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.161622279206714e-06,
+      "loss": 1.9564,
+      "step": 10350
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.157908074813504e-06,
+      "loss": 1.9597,
+      "step": 10355
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.154193783195575e-06,
+      "loss": 1.9684,
+      "step": 10360
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1504794064046184e-06,
+      "loss": 1.9497,
+      "step": 10365
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.146764946492368e-06,
+      "loss": 1.9517,
+      "step": 10370
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.143050405510601e-06,
+      "loss": 1.9475,
+      "step": 10375
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.139335785511147e-06,
+      "loss": 1.9556,
+      "step": 10380
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.135621088545874e-06,
+      "loss": 1.9799,
+      "step": 10385
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.131906316666691e-06,
+      "loss": 1.942,
+      "step": 10390
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1281914719255524e-06,
+      "loss": 1.976,
+      "step": 10395
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.12447655637445e-06,
+      "loss": 1.9535,
+      "step": 10400
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.12076157206542e-06,
+      "loss": 1.9492,
+      "step": 10405
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.11704652105053e-06,
+      "loss": 1.9594,
+      "step": 10410
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.113331405381884e-06,
+      "loss": 1.991,
+      "step": 10415
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.10961622711163e-06,
+      "loss": 1.9971,
+      "step": 10420
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.105900988291943e-06,
+      "loss": 1.969,
+      "step": 10425
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.102185690975034e-06,
+      "loss": 1.9781,
+      "step": 10430
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.098470337213145e-06,
+      "loss": 1.9659,
+      "step": 10435
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.094754929058553e-06,
+      "loss": 1.9488,
+      "step": 10440
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.091039468563559e-06,
+      "loss": 1.9949,
+      "step": 10445
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.087323957780499e-06,
+      "loss": 1.9516,
+      "step": 10450
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.083608398761733e-06,
+      "loss": 1.9839,
+      "step": 10455
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.079892793559647e-06,
+      "loss": 1.9571,
+      "step": 10460
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0761771442266574e-06,
+      "loss": 1.965,
+      "step": 10465
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.072461452815199e-06,
+      "loss": 1.9342,
+      "step": 10470
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.068745721377734e-06,
+      "loss": 1.9472,
+      "step": 10475
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.065029951966744e-06,
+      "loss": 1.9686,
+      "step": 10480
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.061314146634735e-06,
+      "loss": 1.98,
+      "step": 10485
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.057598307434229e-06,
+      "loss": 1.9619,
+      "step": 10490
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.053882436417769e-06,
+      "loss": 1.971,
+      "step": 10495
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.050166535637914e-06,
+      "loss": 1.9436,
+      "step": 10500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.046450607147242e-06,
+      "loss": 1.9535,
+      "step": 10505
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.042734652998344e-06,
+      "loss": 1.9586,
+      "step": 10510
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.039018675243827e-06,
+      "loss": 1.9532,
+      "step": 10515
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.035302675936309e-06,
+      "loss": 1.9721,
+      "step": 10520
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.03158665712842e-06,
+      "loss": 1.957,
+      "step": 10525
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0278706208728026e-06,
+      "loss": 1.9931,
+      "step": 10530
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.024154569222107e-06,
+      "loss": 1.9573,
+      "step": 10535
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.020438504228995e-06,
+      "loss": 1.96,
+      "step": 10540
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0167224279461315e-06,
+      "loss": 1.9464,
+      "step": 10545
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.013006342426192e-06,
+      "loss": 1.9388,
+      "step": 10550
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.009290249721854e-06,
+      "loss": 1.9656,
+      "step": 10555
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.005574151885801e-06,
+      "loss": 1.9538,
+      "step": 10560
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.001858050970716e-06,
+      "loss": 1.9476,
+      "step": 10565
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.998141949029287e-06,
+      "loss": 1.9521,
+      "step": 10570
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.994425848114201e-06,
+      "loss": 1.9389,
+      "step": 10575
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.990709750278148e-06,
+      "loss": 1.9722,
+      "step": 10580
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.986993657573809e-06,
+      "loss": 1.9543,
+      "step": 10585
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.983277572053869e-06,
+      "loss": 1.972,
+      "step": 10590
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.979561495771006e-06,
+      "loss": 1.9366,
+      "step": 10595
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9758454307778944e-06,
+      "loss": 1.9914,
+      "step": 10600
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9721293791272e-06,
+      "loss": 1.9429,
+      "step": 10605
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.968413342871582e-06,
+      "loss": 1.9898,
+      "step": 10610
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.964697324063694e-06,
+      "loss": 1.9753,
+      "step": 10615
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.960981324756174e-06,
+      "loss": 1.9461,
+      "step": 10620
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.957265347001655e-06,
+      "loss": 1.9424,
+      "step": 10625
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.953549392852759e-06,
+      "loss": 1.9768,
+      "step": 10630
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.949833464362087e-06,
+      "loss": 1.9474,
+      "step": 10635
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.946117563582234e-06,
+      "loss": 1.9448,
+      "step": 10640
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.942401692565772e-06,
+      "loss": 1.9679,
+      "step": 10645
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.938685853365267e-06,
+      "loss": 1.9336,
+      "step": 10650
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.934970048033257e-06,
+      "loss": 1.9595,
+      "step": 10655
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.931254278622267e-06,
+      "loss": 1.9635,
+      "step": 10660
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.927538547184802e-06,
+      "loss": 1.9382,
+      "step": 10665
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.923822855773343e-06,
+      "loss": 1.9451,
+      "step": 10670
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9201072064403544e-06,
+      "loss": 1.9594,
+      "step": 10675
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.916391601238268e-06,
+      "loss": 1.975,
+      "step": 10680
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.912676042219503e-06,
+      "loss": 1.9555,
+      "step": 10685
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.908960531436442e-06,
+      "loss": 1.9514,
+      "step": 10690
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.905245070941448e-06,
+      "loss": 1.9586,
+      "step": 10695
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.901529662786856e-06,
+      "loss": 1.9479,
+      "step": 10700
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.897814309024967e-06,
+      "loss": 1.9428,
+      "step": 10705
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.89409901170806e-06,
+      "loss": 1.9473,
+      "step": 10710
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.890383772888371e-06,
+      "loss": 1.9548,
+      "step": 10715
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.886668594618118e-06,
+      "loss": 1.9431,
+      "step": 10720
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.882953478949473e-06,
+      "loss": 1.9615,
+      "step": 10725
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.879238427934582e-06,
+      "loss": 1.951,
+      "step": 10730
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.875523443625551e-06,
+      "loss": 1.9551,
+      "step": 10735
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.871808528074449e-06,
+      "loss": 1.9715,
+      "step": 10740
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.868093683333312e-06,
+      "loss": 1.9626,
+      "step": 10745
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.864378911454128e-06,
+      "loss": 1.982,
+      "step": 10750
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.860664214488856e-06,
+      "loss": 1.9601,
+      "step": 10755
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8569495944894005e-06,
+      "loss": 1.9546,
+      "step": 10760
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.853235053507633e-06,
+      "loss": 1.9385,
+      "step": 10765
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.849520593595383e-06,
+      "loss": 1.9698,
+      "step": 10770
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8458062168044254e-06,
+      "loss": 1.9806,
+      "step": 10775
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.842091925186498e-06,
+      "loss": 1.9549,
+      "step": 10780
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.838377720793288e-06,
+      "loss": 1.9592,
+      "step": 10785
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.834663605676433e-06,
+      "loss": 1.9478,
+      "step": 10790
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8309495818875245e-06,
+      "loss": 1.945,
+      "step": 10795
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.827235651478102e-06,
+      "loss": 1.9392,
+      "step": 10800
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.823521816499655e-06,
+      "loss": 1.9311,
+      "step": 10805
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.819808079003615e-06,
+      "loss": 1.9598,
+      "step": 10810
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.816094441041368e-06,
+      "loss": 1.9934,
+      "step": 10815
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.812380904664237e-06,
+      "loss": 1.9725,
+      "step": 10820
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.808667471923496e-06,
+      "loss": 1.9738,
+      "step": 10825
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.804954144870353e-06,
+      "loss": 1.9609,
+      "step": 10830
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.801240925555966e-06,
+      "loss": 1.9458,
+      "step": 10835
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.797527816031432e-06,
+      "loss": 1.9635,
+      "step": 10840
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.7938148183477815e-06,
+      "loss": 1.9472,
+      "step": 10845
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.7901019345559904e-06,
+      "loss": 1.988,
+      "step": 10850
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.786389166706967e-06,
+      "loss": 1.9686,
+      "step": 10855
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.78267651685156e-06,
+      "loss": 1.939,
+      "step": 10860
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.778963987040544e-06,
+      "loss": 1.959,
+      "step": 10865
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.775251579324639e-06,
+      "loss": 1.9652,
+      "step": 10870
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.771539295754493e-06,
+      "loss": 1.9671,
+      "step": 10875
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.767827138380678e-06,
+      "loss": 1.9583,
+      "step": 10880
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.764115109253709e-06,
+      "loss": 1.9673,
+      "step": 10885
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7604032104240185e-06,
+      "loss": 1.9361,
+      "step": 10890
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.756691443941976e-06,
+      "loss": 1.9509,
+      "step": 10895
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.752979811857874e-06,
+      "loss": 2.0083,
+      "step": 10900
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.749268316221929e-06,
+      "loss": 1.9481,
+      "step": 10905
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.745556959084286e-06,
+      "loss": 1.9279,
+      "step": 10910
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.74184574249501e-06,
+      "loss": 1.9583,
+      "step": 10915
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.738134668504093e-06,
+      "loss": 1.9972,
+      "step": 10920
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.734423739161441e-06,
+      "loss": 1.9537,
+      "step": 10925
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.730712956516887e-06,
+      "loss": 1.9441,
+      "step": 10930
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.727002322620182e-06,
+      "loss": 1.9621,
+      "step": 10935
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.723291839520988e-06,
+      "loss": 1.9512,
+      "step": 10940
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.719581509268894e-06,
+      "loss": 1.9473,
+      "step": 10945
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.715871333913396e-06,
+      "loss": 1.9476,
+      "step": 10950
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.712161315503912e-06,
+      "loss": 1.9548,
+      "step": 10955
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.708451456089762e-06,
+      "loss": 1.9423,
+      "step": 10960
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.70474175772019e-06,
+      "loss": 1.9553,
+      "step": 10965
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.701032222444348e-06,
+      "loss": 1.9787,
+      "step": 10970
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.697322852311293e-06,
+      "loss": 1.9708,
+      "step": 10975
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.693613649369996e-06,
+      "loss": 1.9478,
+      "step": 10980
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.689904615669332e-06,
+      "loss": 1.9229,
+      "step": 10985
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.686195753258088e-06,
+      "loss": 1.9512,
+      "step": 10990
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.682487064184946e-06,
+      "loss": 1.9628,
+      "step": 10995
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.678778550498504e-06,
+      "loss": 1.9522,
+      "step": 11000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.675070214247258e-06,
+      "loss": 1.9501,
+      "step": 11005
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6713620574796025e-06,
+      "loss": 1.941,
+      "step": 11010
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.667654082243841e-06,
+      "loss": 1.9638,
+      "step": 11015
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.663946290588167e-06,
+      "loss": 1.9528,
+      "step": 11020
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.660238684560681e-06,
+      "loss": 1.9674,
+      "step": 11025
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.656531266209377e-06,
+      "loss": 1.9459,
+      "step": 11030
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.652824037582145e-06,
+      "loss": 1.9559,
+      "step": 11035
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.649117000726773e-06,
+      "loss": 1.9608,
+      "step": 11040
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.645410157690938e-06,
+      "loss": 1.9503,
+      "step": 11045
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.641703510522218e-06,
+      "loss": 1.956,
+      "step": 11050
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.637997061268072e-06,
+      "loss": 1.9589,
+      "step": 11055
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.634290811975862e-06,
+      "loss": 1.9668,
+      "step": 11060
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6305847646928285e-06,
+      "loss": 1.9523,
+      "step": 11065
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.626878921466105e-06,
+      "loss": 1.9501,
+      "step": 11070
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6231732843427164e-06,
+      "loss": 1.9779,
+      "step": 11075
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.619467855369566e-06,
+      "loss": 1.9637,
+      "step": 11080
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6157626365934486e-06,
+      "loss": 1.9505,
+      "step": 11085
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.612057630061037e-06,
+      "loss": 1.9449,
+      "step": 11090
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6083528378188925e-06,
+      "loss": 1.9799,
+      "step": 11095
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6046482619134534e-06,
+      "loss": 1.9596,
+      "step": 11100
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.600943904391041e-06,
+      "loss": 1.955,
+      "step": 11105
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.597239767297859e-06,
+      "loss": 1.96,
+      "step": 11110
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.5935358526799804e-06,
+      "loss": 1.9477,
+      "step": 11115
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.589832162583366e-06,
+      "loss": 1.9614,
+      "step": 11120
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.5861286990538425e-06,
+      "loss": 1.9739,
+      "step": 11125
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.582425464137121e-06,
+      "loss": 1.9493,
+      "step": 11130
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.578722459878776e-06,
+      "loss": 1.9911,
+      "step": 11135
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.575019688324264e-06,
+      "loss": 1.9715,
+      "step": 11140
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.571317151518911e-06,
+      "loss": 1.961,
+      "step": 11145
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.567614851507906e-06,
+      "loss": 1.9265,
+      "step": 11150
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.5639127903363175e-06,
+      "loss": 1.9679,
+      "step": 11155
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.5602109700490735e-06,
+      "loss": 1.9582,
+      "step": 11160
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.556509392690978e-06,
+      "loss": 1.9696,
+      "step": 11165
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.552808060306689e-06,
+      "loss": 1.9831,
+      "step": 11170
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.54910697494074e-06,
+      "loss": 1.9634,
+      "step": 11175
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.545406138637523e-06,
+      "loss": 1.9509,
+      "step": 11180
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.541705553441292e-06,
+      "loss": 1.9678,
+      "step": 11185
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.538005221396166e-06,
+      "loss": 1.954,
+      "step": 11190
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.534305144546118e-06,
+      "loss": 1.9525,
+      "step": 11195
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.530605324934987e-06,
+      "loss": 1.9504,
+      "step": 11200
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.526905764606464e-06,
+      "loss": 1.9574,
+      "step": 11205
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.523206465604099e-06,
+      "loss": 1.954,
+      "step": 11210
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.519507429971301e-06,
+      "loss": 1.9584,
+      "step": 11215
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.51580865975133e-06,
+      "loss": 1.9733,
+      "step": 11220
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.512110156987299e-06,
+      "loss": 1.9582,
+      "step": 11225
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.508411923722172e-06,
+      "loss": 1.9527,
+      "step": 11230
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.504713961998771e-06,
+      "loss": 1.9667,
+      "step": 11235
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.501016273859758e-06,
+      "loss": 1.9493,
+      "step": 11240
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.497318861347653e-06,
+      "loss": 1.9635,
+      "step": 11245
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.49362172650482e-06,
+      "loss": 1.9452,
+      "step": 11250
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.489924871373466e-06,
+      "loss": 1.983,
+      "step": 11255
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.4862282979956525e-06,
+      "loss": 1.9699,
+      "step": 11260
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.482532008413275e-06,
+      "loss": 1.9568,
+      "step": 11265
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.47883600466808e-06,
+      "loss": 1.9458,
+      "step": 11270
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.4751402888016525e-06,
+      "loss": 1.9432,
+      "step": 11275
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.471444862855418e-06,
+      "loss": 1.9522,
+      "step": 11280
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.467749728870645e-06,
+      "loss": 1.9635,
+      "step": 11285
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.464054888888438e-06,
+      "loss": 1.9689,
+      "step": 11290
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.4603603449497414e-06,
+      "loss": 1.9895,
+      "step": 11295
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.456666099095332e-06,
+      "loss": 1.9556,
+      "step": 11300
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.452972153365827e-06,
+      "loss": 1.9653,
+      "step": 11305
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.449278509801672e-06,
+      "loss": 1.957,
+      "step": 11310
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.445585170443152e-06,
+      "loss": 1.9442,
+      "step": 11315
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.441892137330382e-06,
+      "loss": 1.9548,
+      "step": 11320
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.438199412503304e-06,
+      "loss": 1.978,
+      "step": 11325
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.434506998001696e-06,
+      "loss": 1.9437,
+      "step": 11330
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4308148958651556e-06,
+      "loss": 1.9475,
+      "step": 11335
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.427123108133119e-06,
+      "loss": 1.937,
+      "step": 11340
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.423431636844839e-06,
+      "loss": 1.9276,
+      "step": 11345
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4197404840394015e-06,
+      "loss": 1.9308,
+      "step": 11350
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4160496517557105e-06,
+      "loss": 1.9479,
+      "step": 11355
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.412359142032494e-06,
+      "loss": 1.9438,
+      "step": 11360
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4086689569083084e-06,
+      "loss": 1.9653,
+      "step": 11365
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.404979098421519e-06,
+      "loss": 1.9554,
+      "step": 11370
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.401289568610321e-06,
+      "loss": 1.9458,
+      "step": 11375
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.397600369512722e-06,
+      "loss": 1.9252,
+      "step": 11380
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.393911503166548e-06,
+      "loss": 1.9464,
+      "step": 11385
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.390222971609448e-06,
+      "loss": 1.9654,
+      "step": 11390
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.386534776878874e-06,
+      "loss": 1.9512,
+      "step": 11395
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.3828469210121025e-06,
+      "loss": 1.9951,
+      "step": 11400
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.379159406046215e-06,
+      "loss": 1.9515,
+      "step": 11405
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.375472234018112e-06,
+      "loss": 1.97,
+      "step": 11410
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.371785406964497e-06,
+      "loss": 1.9858,
+      "step": 11415
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.368098926921889e-06,
+      "loss": 1.9414,
+      "step": 11420
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.364412795926615e-06,
+      "loss": 1.9524,
+      "step": 11425
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.360727016014803e-06,
+      "loss": 1.9556,
+      "step": 11430
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.357041589222396e-06,
+      "loss": 1.9732,
+      "step": 11435
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.353356517585132e-06,
+      "loss": 1.9387,
+      "step": 11440
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.349671803138564e-06,
+      "loss": 1.9621,
+      "step": 11445
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.345987447918037e-06,
+      "loss": 1.9636,
+      "step": 11450
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.342303453958703e-06,
+      "loss": 1.9452,
+      "step": 11455
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.338619823295517e-06,
+      "loss": 1.9151,
+      "step": 11460
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.334936557963228e-06,
+      "loss": 1.9445,
+      "step": 11465
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.331253659996385e-06,
+      "loss": 1.9563,
+      "step": 11470
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.327571131429336e-06,
+      "loss": 1.9439,
+      "step": 11475
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.323888974296223e-06,
+      "loss": 1.938,
+      "step": 11480
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.320207190630982e-06,
+      "loss": 1.953,
+      "step": 11485
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.3165257824673435e-06,
+      "loss": 1.9682,
+      "step": 11490
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.312844751838835e-06,
+      "loss": 1.9412,
+      "step": 11495
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.309164100778766e-06,
+      "loss": 1.9235,
+      "step": 11500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.305483831320248e-06,
+      "loss": 1.9422,
+      "step": 11505
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.301803945496169e-06,
+      "loss": 1.933,
+      "step": 11510
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.2981244453392156e-06,
+      "loss": 1.9494,
+      "step": 11515
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.294445332881857e-06,
+      "loss": 1.9326,
+      "step": 11520
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.290766610156347e-06,
+      "loss": 1.953,
+      "step": 11525
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.287088279194727e-06,
+      "loss": 1.9347,
+      "step": 11530
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.283410342028821e-06,
+      "loss": 1.9498,
+      "step": 11535
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.279732800690237e-06,
+      "loss": 1.9521,
+      "step": 11540
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.276055657210357e-06,
+      "loss": 1.9575,
+      "step": 11545
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.272378913620356e-06,
+      "loss": 1.977,
+      "step": 11550
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.2687025719511735e-06,
+      "loss": 1.9719,
+      "step": 11555
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.2650266342335385e-06,
+      "loss": 1.9352,
+      "step": 11560
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.261351102497956e-06,
+      "loss": 1.9431,
+      "step": 11565
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.257675978774696e-06,
+      "loss": 1.9292,
+      "step": 11570
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.254001265093817e-06,
+      "loss": 1.9395,
+      "step": 11575
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.25032696348514e-06,
+      "loss": 1.9483,
+      "step": 11580
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.246653075978265e-06,
+      "loss": 1.9534,
+      "step": 11585
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.242979604602563e-06,
+      "loss": 1.939,
+      "step": 11590
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.239306551387172e-06,
+      "loss": 1.9583,
+      "step": 11595
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.2356339183610005e-06,
+      "loss": 1.9548,
+      "step": 11600
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.231961707552723e-06,
+      "loss": 1.946,
+      "step": 11605
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.228289920990788e-06,
+      "loss": 1.9791,
+      "step": 11610
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.224618560703397e-06,
+      "loss": 1.9686,
+      "step": 11615
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.220947628718527e-06,
+      "loss": 1.9486,
+      "step": 11620
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.217277127063918e-06,
+      "loss": 1.9409,
+      "step": 11625
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.213607057767062e-06,
+      "loss": 1.9315,
+      "step": 11630
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.209937422855226e-06,
+      "loss": 1.9314,
+      "step": 11635
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.206268224355425e-06,
+      "loss": 1.946,
+      "step": 11640
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.202599464294441e-06,
+      "loss": 1.9515,
+      "step": 11645
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.19893114469881e-06,
+      "loss": 1.9453,
+      "step": 11650
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.195263267594826e-06,
+      "loss": 1.9522,
+      "step": 11655
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.191595835008538e-06,
+      "loss": 1.9501,
+      "step": 11660
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.18792884896575e-06,
+      "loss": 1.9455,
+      "step": 11665
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.184262311492022e-06,
+      "loss": 1.9537,
+      "step": 11670
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.180596224612657e-06,
+      "loss": 1.9328,
+      "step": 11675
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.1769305903527214e-06,
+      "loss": 1.9409,
+      "step": 11680
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.173265410737021e-06,
+      "loss": 1.9404,
+      "step": 11685
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.169600687790118e-06,
+      "loss": 1.9517,
+      "step": 11690
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.165936423536319e-06,
+      "loss": 1.956,
+      "step": 11695
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.162272619999676e-06,
+      "loss": 1.9503,
+      "step": 11700
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.15860927920399e-06,
+      "loss": 1.9703,
+      "step": 11705
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.1549464031728046e-06,
+      "loss": 1.9639,
+      "step": 11710
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.151283993929405e-06,
+      "loss": 1.9324,
+      "step": 11715
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.14762205349682e-06,
+      "loss": 1.9585,
+      "step": 11720
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.1439605838978205e-06,
+      "loss": 1.9331,
+      "step": 11725
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.140299587154919e-06,
+      "loss": 1.9617,
+      "step": 11730
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.1366390652903596e-06,
+      "loss": 1.9592,
+      "step": 11735
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.132979020326134e-06,
+      "loss": 1.9366,
+      "step": 11740
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.129319454283958e-06,
+      "loss": 1.9521,
+      "step": 11745
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.125660369185297e-06,
+      "loss": 1.9379,
+      "step": 11750
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.122001767051339e-06,
+      "loss": 1.9386,
+      "step": 11755
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.118343649903012e-06,
+      "loss": 1.9598,
+      "step": 11760
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.114686019760976e-06,
+      "loss": 1.9458,
+      "step": 11765
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.111028878645616e-06,
+      "loss": 1.9519,
+      "step": 11770
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.107372228577054e-06,
+      "loss": 1.9746,
+      "step": 11775
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.103716071575136e-06,
+      "loss": 1.9688,
+      "step": 11780
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.100060409659441e-06,
+      "loss": 1.9494,
+      "step": 11785
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.096405244849265e-06,
+      "loss": 1.954,
+      "step": 11790
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.092750579163639e-06,
+      "loss": 1.9331,
+      "step": 11795
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.089096414621317e-06,
+      "loss": 1.9582,
+      "step": 11800
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.085442753240767e-06,
+      "loss": 1.9645,
+      "step": 11805
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0817895970401924e-06,
+      "loss": 1.9608,
+      "step": 11810
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.078136948037506e-06,
+      "loss": 1.9368,
+      "step": 11815
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.07448480825035e-06,
+      "loss": 1.9307,
+      "step": 11820
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.070833179696077e-06,
+      "loss": 1.9454,
+      "step": 11825
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.067182064391762e-06,
+      "loss": 1.9421,
+      "step": 11830
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.063531464354197e-06,
+      "loss": 1.9767,
+      "step": 11835
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0598813815998865e-06,
+      "loss": 1.9781,
+      "step": 11840
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.056231818145051e-06,
+      "loss": 1.9502,
+      "step": 11845
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.052582776005623e-06,
+      "loss": 1.9405,
+      "step": 11850
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.04893425719725e-06,
+      "loss": 1.9654,
+      "step": 11855
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.045286263735284e-06,
+      "loss": 1.9364,
+      "step": 11860
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.041638797634795e-06,
+      "loss": 1.9393,
+      "step": 11865
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.037991860910558e-06,
+      "loss": 1.938,
+      "step": 11870
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.03434545557705e-06,
+      "loss": 1.9553,
+      "step": 11875
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.030699583648467e-06,
+      "loss": 1.9447,
+      "step": 11880
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.027054247138698e-06,
+      "loss": 1.9592,
+      "step": 11885
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.023409448061343e-06,
+      "loss": 1.9689,
+      "step": 11890
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.019765188429704e-06,
+      "loss": 1.9508,
+      "step": 11895
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.016121470256783e-06,
+      "loss": 1.9401,
+      "step": 11900
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.012478295555287e-06,
+      "loss": 1.9459,
+      "step": 11905
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.008835666337616e-06,
+      "loss": 1.9323,
+      "step": 11910
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.00519358461588e-06,
+      "loss": 1.9335,
+      "step": 11915
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0015520524018715e-06,
+      "loss": 1.947,
+      "step": 11920
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.997911071707094e-06,
+      "loss": 1.9468,
+      "step": 11925
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.9942706445427345e-06,
+      "loss": 1.9548,
+      "step": 11930
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.990630772919683e-06,
+      "loss": 1.9641,
+      "step": 11935
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.98699145884852e-06,
+      "loss": 1.9452,
+      "step": 11940
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.983352704339514e-06,
+      "loss": 1.9475,
+      "step": 11945
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.97971451140263e-06,
+      "loss": 1.9491,
+      "step": 11950
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.976076882047519e-06,
+      "loss": 1.9279,
+      "step": 11955
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.972439818283523e-06,
+      "loss": 1.9526,
+      "step": 11960
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.96880332211967e-06,
+      "loss": 1.9492,
+      "step": 11965
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9651673955646755e-06,
+      "loss": 1.9497,
+      "step": 11970
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.961532040626942e-06,
+      "loss": 1.9435,
+      "step": 11975
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.95789725931455e-06,
+      "loss": 1.9372,
+      "step": 11980
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.954263053635274e-06,
+      "loss": 1.9322,
+      "step": 11985
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.950629425596556e-06,
+      "loss": 1.9437,
+      "step": 11990
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.946996377205534e-06,
+      "loss": 1.9447,
+      "step": 11995
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.943363910469014e-06,
+      "loss": 1.9322,
+      "step": 12000
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 21135,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "total_flos": 2.010023756562432e+16,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}