diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,23838 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.0,
+  "eval_steps": 500,
+  "global_step": 19818,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.5317860746720484e-08,
+      "loss": 1.9117,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.765893037336024e-07,
+      "loss": 1.8639,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.531786074672048e-07,
+      "loss": 1.8303,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.297679112008072e-07,
+      "loss": 1.8892,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.063572149344096e-07,
+      "loss": 1.8748,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.82946518668012e-07,
+      "loss": 1.8591,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0595358224016144e-06,
+      "loss": 1.8369,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2361251261352167e-06,
+      "loss": 1.826,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4127144298688192e-06,
+      "loss": 1.8198,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5893037336024215e-06,
+      "loss": 1.8308,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.765893037336024e-06,
+      "loss": 1.8138,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9424823410696264e-06,
+      "loss": 1.8355,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.119071644803229e-06,
+      "loss": 1.8033,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2956609485368314e-06,
+      "loss": 1.7987,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4722502522704335e-06,
+      "loss": 1.7345,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.648839556004036e-06,
+      "loss": 1.739,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8254288597376385e-06,
+      "loss": 1.7845,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.002018163471241e-06,
+      "loss": 1.7468,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.178607467204843e-06,
+      "loss": 1.7101,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.355196770938446e-06,
+      "loss": 1.6895,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.531786074672048e-06,
+      "loss": 1.6488,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7083753784056506e-06,
+      "loss": 1.6526,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.884964682139253e-06,
+      "loss": 1.6533,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.061553985872855e-06,
+      "loss": 1.655,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.238143289606458e-06,
+      "loss": 1.6066,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.41473259334006e-06,
+      "loss": 1.5711,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.591321897073663e-06,
+      "loss": 1.5883,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.767911200807264e-06,
+      "loss": 1.5776,
+      "step": 135
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.944500504540867e-06,
+      "loss": 1.5862,
+      "step": 140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.12108980827447e-06,
+      "loss": 1.575,
+      "step": 145
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.297679112008072e-06,
+      "loss": 1.541,
+      "step": 150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.4742684157416745e-06,
+      "loss": 1.5497,
+      "step": 155
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.650857719475277e-06,
+      "loss": 1.5478,
+      "step": 160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.8274470232088795e-06,
+      "loss": 1.5249,
+      "step": 165
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.004036326942482e-06,
+      "loss": 1.5566,
+      "step": 170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.1806256306760845e-06,
+      "loss": 1.5126,
+      "step": 175
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.357214934409686e-06,
+      "loss": 1.5391,
+      "step": 180
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.533804238143289e-06,
+      "loss": 1.5098,
+      "step": 185
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.710393541876892e-06,
+      "loss": 1.5097,
+      "step": 190
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.886982845610494e-06,
+      "loss": 1.5143,
+      "step": 195
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.063572149344096e-06,
+      "loss": 1.5138,
+      "step": 200
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.240161453077699e-06,
+      "loss": 1.4911,
+      "step": 205
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.416750756811301e-06,
+      "loss": 1.5041,
+      "step": 210
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.593340060544904e-06,
+      "loss": 1.4957,
+      "step": 215
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.769929364278505e-06,
+      "loss": 1.4993,
+      "step": 220
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.946518668012109e-06,
+      "loss": 1.5228,
+      "step": 225
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.12310797174571e-06,
+      "loss": 1.4771,
+      "step": 230
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.299697275479312e-06,
+      "loss": 1.5206,
+      "step": 235
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.476286579212915e-06,
+      "loss": 1.5096,
+      "step": 240
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.652875882946519e-06,
+      "loss": 1.4673,
+      "step": 245
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.82946518668012e-06,
+      "loss": 1.4601,
+      "step": 250
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.006054490413722e-06,
+      "loss": 1.4597,
+      "step": 255
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.182643794147326e-06,
+      "loss": 1.461,
+      "step": 260
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.359233097880929e-06,
+      "loss": 1.4755,
+      "step": 265
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.535822401614529e-06,
+      "loss": 1.4568,
+      "step": 270
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.712411705348132e-06,
+      "loss": 1.4599,
+      "step": 275
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.889001009081734e-06,
+      "loss": 1.4689,
+      "step": 280
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.0065590312815337e-05,
+      "loss": 1.4647,
+      "step": 285
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.024217961654894e-05,
+      "loss": 1.4505,
+      "step": 290
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.0418768920282542e-05,
+      "loss": 1.4633,
+      "step": 295
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0595358224016144e-05,
+      "loss": 1.429,
+      "step": 300
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0771947527749746e-05,
+      "loss": 1.4354,
+      "step": 305
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0948536831483349e-05,
+      "loss": 1.44,
+      "step": 310
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1125126135216952e-05,
+      "loss": 1.4193,
+      "step": 315
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1301715438950554e-05,
+      "loss": 1.4237,
+      "step": 320
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1478304742684157e-05,
+      "loss": 1.4436,
+      "step": 325
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1654894046417759e-05,
+      "loss": 1.4502,
+      "step": 330
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.183148335015136e-05,
+      "loss": 1.4329,
+      "step": 335
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2008072653884964e-05,
+      "loss": 1.4436,
+      "step": 340
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2184661957618566e-05,
+      "loss": 1.4299,
+      "step": 345
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2361251261352169e-05,
+      "loss": 1.4459,
+      "step": 350
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.253784056508577e-05,
+      "loss": 1.4425,
+      "step": 355
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2714429868819372e-05,
+      "loss": 1.4678,
+      "step": 360
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2891019172552976e-05,
+      "loss": 1.4184,
+      "step": 365
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3067608476286577e-05,
+      "loss": 1.4292,
+      "step": 370
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.324419778002018e-05,
+      "loss": 1.4019,
+      "step": 375
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3420787083753784e-05,
+      "loss": 1.4262,
+      "step": 380
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3597376387487384e-05,
+      "loss": 1.417,
+      "step": 385
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3773965691220987e-05,
+      "loss": 1.3865,
+      "step": 390
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.395055499495459e-05,
+      "loss": 1.421,
+      "step": 395
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4127144298688192e-05,
+      "loss": 1.4049,
+      "step": 400
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4303733602421796e-05,
+      "loss": 1.4078,
+      "step": 405
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4480322906155397e-05,
+      "loss": 1.4309,
+      "step": 410
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4656912209888999e-05,
+      "loss": 1.3779,
+      "step": 415
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4833501513622602e-05,
+      "loss": 1.4095,
+      "step": 420
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.5010090817356204e-05,
+      "loss": 1.3682,
+      "step": 425
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5186680121089808e-05,
+      "loss": 1.3923,
+      "step": 430
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.536326942482341e-05,
+      "loss": 1.4021,
+      "step": 435
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.553985872855701e-05,
+      "loss": 1.3986,
+      "step": 440
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5716448032290613e-05,
+      "loss": 1.3776,
+      "step": 445
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5893037336024218e-05,
+      "loss": 1.387,
+      "step": 450
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.606962663975782e-05,
+      "loss": 1.3973,
+      "step": 455
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.624621594349142e-05,
+      "loss": 1.4077,
+      "step": 460
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6422805247225026e-05,
+      "loss": 1.3865,
+      "step": 465
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6599394550958624e-05,
+      "loss": 1.3505,
+      "step": 470
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.677598385469223e-05,
+      "loss": 1.4001,
+      "step": 475
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.695257315842583e-05,
+      "loss": 1.3902,
+      "step": 480
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.7129162462159433e-05,
+      "loss": 1.3664,
+      "step": 485
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.7305751765893038e-05,
+      "loss": 1.3792,
+      "step": 490
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.748234106962664e-05,
+      "loss": 1.3861,
+      "step": 495
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.765893037336024e-05,
+      "loss": 1.3652,
+      "step": 500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7835519677093846e-05,
+      "loss": 1.3422,
+      "step": 505
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8012108980827444e-05,
+      "loss": 1.3769,
+      "step": 510
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8188698284561046e-05,
+      "loss": 1.3347,
+      "step": 515
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.836528758829465e-05,
+      "loss": 1.3645,
+      "step": 520
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8541876892028253e-05,
+      "loss": 1.3784,
+      "step": 525
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8718466195761858e-05,
+      "loss": 1.3919,
+      "step": 530
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8895055499495456e-05,
+      "loss": 1.376,
+      "step": 535
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9071644803229058e-05,
+      "loss": 1.3446,
+      "step": 540
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9248234106962663e-05,
+      "loss": 1.3572,
+      "step": 545
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9424823410696264e-05,
+      "loss": 1.3322,
+      "step": 550
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.960141271442987e-05,
+      "loss": 1.3766,
+      "step": 555
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9778002018163468e-05,
+      "loss": 1.3636,
+      "step": 560
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9954591321897073e-05,
+      "loss": 1.3218,
+      "step": 565
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.0131180625630674e-05,
+      "loss": 1.3119,
+      "step": 570
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.0307769929364276e-05,
+      "loss": 1.3491,
+      "step": 575
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.048435923309788e-05,
+      "loss": 1.36,
+      "step": 580
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.066094853683148e-05,
+      "loss": 1.3514,
+      "step": 585
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.0837537840565084e-05,
+      "loss": 1.3434,
+      "step": 590
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1014127144298686e-05,
+      "loss": 1.3552,
+      "step": 595
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1190716448032288e-05,
+      "loss": 1.3345,
+      "step": 600
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1367305751765893e-05,
+      "loss": 1.3232,
+      "step": 605
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.154389505549949e-05,
+      "loss": 1.3561,
+      "step": 610
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1720484359233096e-05,
+      "loss": 1.331,
+      "step": 615
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1897073662966698e-05,
+      "loss": 1.3599,
+      "step": 620
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.20736629667003e-05,
+      "loss": 1.3184,
+      "step": 625
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.2250252270433905e-05,
+      "loss": 1.3405,
+      "step": 630
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.2426841574167503e-05,
+      "loss": 1.3293,
+      "step": 635
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.2603430877901108e-05,
+      "loss": 1.3575,
+      "step": 640
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.278002018163471e-05,
+      "loss": 1.3288,
+      "step": 645
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.2956609485368315e-05,
+      "loss": 1.3399,
+      "step": 650
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.3133198789101916e-05,
+      "loss": 1.3364,
+      "step": 655
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.3309788092835518e-05,
+      "loss": 1.311,
+      "step": 660
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.348637739656912e-05,
+      "loss": 1.374,
+      "step": 665
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.366296670030272e-05,
+      "loss": 1.3155,
+      "step": 670
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.3839556004036326e-05,
+      "loss": 1.3295,
+      "step": 675
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.4016145307769928e-05,
+      "loss": 1.3392,
+      "step": 680
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.419273461150353e-05,
+      "loss": 1.3039,
+      "step": 685
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.436932391523713e-05,
+      "loss": 1.3109,
+      "step": 690
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.4545913218970733e-05,
+      "loss": 1.3468,
+      "step": 695
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.4722502522704338e-05,
+      "loss": 1.3249,
+      "step": 700
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.489909182643794e-05,
+      "loss": 1.3326,
+      "step": 705
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.507568113017154e-05,
+      "loss": 1.3159,
+      "step": 710
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.5252270433905143e-05,
+      "loss": 1.3289,
+      "step": 715
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.5428859737638745e-05,
+      "loss": 1.3245,
+      "step": 720
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.560544904137235e-05,
+      "loss": 1.2899,
+      "step": 725
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.578203834510595e-05,
+      "loss": 1.3251,
+      "step": 730
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.5958627648839556e-05,
+      "loss": 1.3565,
+      "step": 735
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.6135216952573155e-05,
+      "loss": 1.3415,
+      "step": 740
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.6311806256306756e-05,
+      "loss": 1.316,
+      "step": 745
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.648839556004036e-05,
+      "loss": 1.3262,
+      "step": 750
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.6664984863773963e-05,
+      "loss": 1.3189,
+      "step": 755
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6841574167507568e-05,
+      "loss": 1.3029,
+      "step": 760
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.701816347124117e-05,
+      "loss": 1.2914,
+      "step": 765
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.7194752774974768e-05,
+      "loss": 1.3675,
+      "step": 770
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.7371342078708373e-05,
+      "loss": 1.3351,
+      "step": 775
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.7547931382441975e-05,
+      "loss": 1.3133,
+      "step": 780
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.772452068617558e-05,
+      "loss": 1.276,
+      "step": 785
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.790110998990918e-05,
+      "loss": 1.2889,
+      "step": 790
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.807769929364278e-05,
+      "loss": 1.3367,
+      "step": 795
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.8254288597376385e-05,
+      "loss": 1.2912,
+      "step": 800
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.8430877901109987e-05,
+      "loss": 1.3215,
+      "step": 805
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.860746720484359e-05,
+      "loss": 1.302,
+      "step": 810
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.8784056508577193e-05,
+      "loss": 1.3101,
+      "step": 815
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.8960645812310795e-05,
+      "loss": 1.3164,
+      "step": 820
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9137235116044397e-05,
+      "loss": 1.2819,
+      "step": 825
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9313824419777998e-05,
+      "loss": 1.312,
+      "step": 830
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9490413723511603e-05,
+      "loss": 1.2903,
+      "step": 835
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9667003027245205e-05,
+      "loss": 1.284,
+      "step": 840
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9843592330978807e-05,
+      "loss": 1.2983,
+      "step": 845
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.002018163471241e-05,
+      "loss": 1.2615,
+      "step": 850
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.019677093844601e-05,
+      "loss": 1.2741,
+      "step": 855
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.0373360242179615e-05,
+      "loss": 1.3109,
+      "step": 860
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.054994954591321e-05,
+      "loss": 1.3325,
+      "step": 865
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.072653884964682e-05,
+      "loss": 1.2764,
+      "step": 870
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.0903128153380423e-05,
+      "loss": 1.2954,
+      "step": 875
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.107971745711402e-05,
+      "loss": 1.2819,
+      "step": 880
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.125630676084763e-05,
+      "loss": 1.3064,
+      "step": 885
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.1432896064581225e-05,
+      "loss": 1.3095,
+      "step": 890
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.160948536831483e-05,
+      "loss": 1.309,
+      "step": 895
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.1786074672048435e-05,
+      "loss": 1.2738,
+      "step": 900
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.196266397578204e-05,
+      "loss": 1.288,
+      "step": 905
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.213925327951564e-05,
+      "loss": 1.2609,
+      "step": 910
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.231584258324924e-05,
+      "loss": 1.298,
+      "step": 915
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.249243188698284e-05,
+      "loss": 1.3077,
+      "step": 920
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.266902119071645e-05,
+      "loss": 1.2627,
+      "step": 925
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.284561049445005e-05,
+      "loss": 1.2545,
+      "step": 930
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.302219979818365e-05,
+      "loss": 1.291,
+      "step": 935
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.319878910191725e-05,
+      "loss": 1.2903,
+      "step": 940
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.3375378405650853e-05,
+      "loss": 1.2931,
+      "step": 945
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.355196770938446e-05,
+      "loss": 1.2236,
+      "step": 950
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.3728557013118064e-05,
+      "loss": 1.2636,
+      "step": 955
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.390514631685166e-05,
+      "loss": 1.2764,
+      "step": 960
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.408173562058526e-05,
+      "loss": 1.2977,
+      "step": 965
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.4258324924318865e-05,
+      "loss": 1.2516,
+      "step": 970
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.443491422805247e-05,
+      "loss": 1.3077,
+      "step": 975
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.4611503531786075e-05,
+      "loss": 1.2606,
+      "step": 980
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.4788092835519674e-05,
+      "loss": 1.2494,
+      "step": 985
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.496468213925328e-05,
+      "loss": 1.2901,
+      "step": 990
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.514127144298688e-05,
+      "loss": 1.2471,
+      "step": 995
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.531786074672048e-05,
+      "loss": 1.2634,
+      "step": 1000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.549445005045408e-05,
+      "loss": 1.2701,
+      "step": 1005
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.567103935418769e-05,
+      "loss": 1.2714,
+      "step": 1010
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.584762865792129e-05,
+      "loss": 1.2761,
+      "step": 1015
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.602421796165489e-05,
+      "loss": 1.271,
+      "step": 1020
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.6200807265388494e-05,
+      "loss": 1.2661,
+      "step": 1025
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.637739656912209e-05,
+      "loss": 1.2629,
+      "step": 1030
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.6553985872855704e-05,
+      "loss": 1.2733,
+      "step": 1035
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.67305751765893e-05,
+      "loss": 1.2535,
+      "step": 1040
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.69071644803229e-05,
+      "loss": 1.2607,
+      "step": 1045
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.7083753784056505e-05,
+      "loss": 1.2801,
+      "step": 1050
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.7260343087790104e-05,
+      "loss": 1.3035,
+      "step": 1055
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.7436932391523715e-05,
+      "loss": 1.2322,
+      "step": 1060
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.7613521695257314e-05,
+      "loss": 1.2636,
+      "step": 1065
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.779011099899091e-05,
+      "loss": 1.2669,
+      "step": 1070
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.796670030272452e-05,
+      "loss": 1.2762,
+      "step": 1075
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8143289606458115e-05,
+      "loss": 1.255,
+      "step": 1080
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.831987891019173e-05,
+      "loss": 1.2812,
+      "step": 1085
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.8496468213925325e-05,
+      "loss": 1.2496,
+      "step": 1090
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.8673057517658924e-05,
+      "loss": 1.2587,
+      "step": 1095
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.884964682139253e-05,
+      "loss": 1.2625,
+      "step": 1100
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.902623612512613e-05,
+      "loss": 1.2686,
+      "step": 1105
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.920282542885974e-05,
+      "loss": 1.2616,
+      "step": 1110
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.937941473259334e-05,
+      "loss": 1.2481,
+      "step": 1115
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.9556004036326935e-05,
+      "loss": 1.2723,
+      "step": 1120
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.973259334006054e-05,
+      "loss": 1.2548,
+      "step": 1125
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.9909182643794146e-05,
+      "loss": 1.2585,
+      "step": 1130
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.008577194752775e-05,
+      "loss": 1.2285,
+      "step": 1135
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.026236125126135e-05,
+      "loss": 1.2481,
+      "step": 1140
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.043895055499495e-05,
+      "loss": 1.2561,
+      "step": 1145
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.061553985872855e-05,
+      "loss": 1.2638,
+      "step": 1150
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.079212916246216e-05,
+      "loss": 1.2517,
+      "step": 1155
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.096871846619576e-05,
+      "loss": 1.2461,
+      "step": 1160
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.114530776992936e-05,
+      "loss": 1.2042,
+      "step": 1165
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.132189707366296e-05,
+      "loss": 1.2494,
+      "step": 1170
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.1498486377396564e-05,
+      "loss": 1.2475,
+      "step": 1175
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.167507568113017e-05,
+      "loss": 1.2195,
+      "step": 1180
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.1851664984863774e-05,
+      "loss": 1.2401,
+      "step": 1185
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.202825428859737e-05,
+      "loss": 1.2886,
+      "step": 1190
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.220484359233097e-05,
+      "loss": 1.2465,
+      "step": 1195
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.2381432896064576e-05,
+      "loss": 1.2466,
+      "step": 1200
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.255802219979818e-05,
+      "loss": 1.2147,
+      "step": 1205
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.2734611503531786e-05,
+      "loss": 1.2504,
+      "step": 1210
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.2911200807265384e-05,
+      "loss": 1.2338,
+      "step": 1215
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.308779011099898e-05,
+      "loss": 1.2168,
+      "step": 1220
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.326437941473259e-05,
+      "loss": 1.2317,
+      "step": 1225
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.344096871846619e-05,
+      "loss": 1.2189,
+      "step": 1230
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.36175580221998e-05,
+      "loss": 1.2742,
+      "step": 1235
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.3794147325933396e-05,
+      "loss": 1.242,
+      "step": 1240
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.3970736629666994e-05,
+      "loss": 1.2723,
+      "step": 1245
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.41473259334006e-05,
+      "loss": 1.2369,
+      "step": 1250
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.4323915237134204e-05,
+      "loss": 1.2599,
+      "step": 1255
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.450050454086781e-05,
+      "loss": 1.2507,
+      "step": 1260
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.467709384460141e-05,
+      "loss": 1.245,
+      "step": 1265
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.4853683148335006e-05,
+      "loss": 1.2099,
+      "step": 1270
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.503027245206861e-05,
+      "loss": 1.2316,
+      "step": 1275
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5206861755802216e-05,
+      "loss": 1.2607,
+      "step": 1280
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.538345105953582e-05,
+      "loss": 1.265,
+      "step": 1285
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.556004036326942e-05,
+      "loss": 1.228,
+      "step": 1290
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.573662966700302e-05,
+      "loss": 1.2748,
+      "step": 1295
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.591321897073663e-05,
+      "loss": 1.1965,
+      "step": 1300
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.608980827447023e-05,
+      "loss": 1.2459,
+      "step": 1305
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.626639757820383e-05,
+      "loss": 1.2948,
+      "step": 1310
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.644298688193743e-05,
+      "loss": 1.2396,
+      "step": 1315
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6619576185671036e-05,
+      "loss": 1.2486,
+      "step": 1320
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.679616548940464e-05,
+      "loss": 1.2599,
+      "step": 1325
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.697275479313824e-05,
+      "loss": 1.2494,
+      "step": 1330
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.7149344096871844e-05,
+      "loss": 1.2509,
+      "step": 1335
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.732593340060544e-05,
+      "loss": 1.2332,
+      "step": 1340
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.750252270433905e-05,
+      "loss": 1.2129,
+      "step": 1345
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.767911200807265e-05,
+      "loss": 1.227,
+      "step": 1350
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.785570131180625e-05,
+      "loss": 1.2339,
+      "step": 1355
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.8032290615539856e-05,
+      "loss": 1.2291,
+      "step": 1360
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.8208879919273454e-05,
+      "loss": 1.23,
+      "step": 1365
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.838546922300706e-05,
+      "loss": 1.2472,
+      "step": 1370
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.8562058526740664e-05,
+      "loss": 1.2474,
+      "step": 1375
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.873864783047426e-05,
+      "loss": 1.2412,
+      "step": 1380
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.891523713420787e-05,
+      "loss": 1.2178,
+      "step": 1385
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9091826437941466e-05,
+      "loss": 1.207,
+      "step": 1390
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.926841574167507e-05,
+      "loss": 1.2189,
+      "step": 1395
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9445005045408676e-05,
+      "loss": 1.192,
+      "step": 1400
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9621594349142274e-05,
+      "loss": 1.2462,
+      "step": 1405
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.979818365287588e-05,
+      "loss": 1.2329,
+      "step": 1410
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.997477295660948e-05,
+      "loss": 1.1895,
+      "step": 1415
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.015136226034308e-05,
+      "loss": 1.2001,
+      "step": 1420
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.032795156407669e-05,
+      "loss": 1.228,
+      "step": 1425
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.0504540867810286e-05,
+      "loss": 1.2446,
+      "step": 1430
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.068113017154389e-05,
+      "loss": 1.2555,
+      "step": 1435
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.085771947527749e-05,
+      "loss": 1.2311,
+      "step": 1440
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.1034308779011094e-05,
+      "loss": 1.225,
+      "step": 1445
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.12108980827447e-05,
+      "loss": 1.2293,
+      "step": 1450
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.13874873864783e-05,
+      "loss": 1.211,
+      "step": 1455
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.15640766902119e-05,
+      "loss": 1.1954,
+      "step": 1460
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.17406659939455e-05,
+      "loss": 1.2281,
+      "step": 1465
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.191725529767911e-05,
+      "loss": 1.2028,
+      "step": 1470
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.209384460141271e-05,
+      "loss": 1.198,
+      "step": 1475
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.227043390514631e-05,
+      "loss": 1.2096,
+      "step": 1480
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.2447023208879915e-05,
+      "loss": 1.2431,
+      "step": 1485
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.262361251261351e-05,
+      "loss": 1.1651,
+      "step": 1490
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.2800201816347125e-05,
+      "loss": 1.1958,
+      "step": 1495
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.297679112008072e-05,
+      "loss": 1.2098,
+      "step": 1500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.315338042381432e-05,
+      "loss": 1.1911,
+      "step": 1505
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.3329969727547926e-05,
+      "loss": 1.2015,
+      "step": 1510
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.3506559031281525e-05,
+      "loss": 1.2086,
+      "step": 1515
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.3683148335015136e-05,
+      "loss": 1.2136,
+      "step": 1520
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.3859737638748735e-05,
+      "loss": 1.2056,
+      "step": 1525
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.403632694248234e-05,
+      "loss": 1.2212,
+      "step": 1530
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.421291624621594e-05,
+      "loss": 1.2036,
+      "step": 1535
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.4389505549949536e-05,
+      "loss": 1.2173,
+      "step": 1540
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.456609485368315e-05,
+      "loss": 1.2216,
+      "step": 1545
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.4742684157416746e-05,
+      "loss": 1.229,
+      "step": 1550
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.491927346115035e-05,
+      "loss": 1.2209,
+      "step": 1555
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.509586276488395e-05,
+      "loss": 1.2682,
+      "step": 1560
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.527245206861755e-05,
+      "loss": 1.1805,
+      "step": 1565
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.544904137235116e-05,
+      "loss": 1.2191,
+      "step": 1570
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.562563067608476e-05,
+      "loss": 1.1704,
+      "step": 1575
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.580221997981836e-05,
+      "loss": 1.2316,
+      "step": 1580
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.597880928355196e-05,
+      "loss": 1.2298,
+      "step": 1585
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.615539858728556e-05,
+      "loss": 1.195,
+      "step": 1590
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.633198789101917e-05,
+      "loss": 1.1971,
+      "step": 1595
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.650857719475277e-05,
+      "loss": 1.2429,
+      "step": 1600
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.6685166498486375e-05,
+      "loss": 1.2021,
+      "step": 1605
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.686175580221997e-05,
+      "loss": 1.2542,
+      "step": 1610
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.703834510595357e-05,
+      "loss": 1.2147,
+      "step": 1615
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.721493440968718e-05,
+      "loss": 1.2487,
+      "step": 1620
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.739152371342078e-05,
+      "loss": 1.2266,
+      "step": 1625
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.7568113017154387e-05,
+      "loss": 1.1644,
+      "step": 1630
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.7744702320887985e-05,
+      "loss": 1.1911,
+      "step": 1635
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.792129162462159e-05,
+      "loss": 1.2173,
+      "step": 1640
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.8097880928355195e-05,
+      "loss": 1.1971,
+      "step": 1645
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.827447023208879e-05,
+      "loss": 1.2038,
+      "step": 1650
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.84510595358224e-05,
+      "loss": 1.2165,
+      "step": 1655
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.8627648839555997e-05,
+      "loss": 1.2141,
+      "step": 1660
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.88042381432896e-05,
+      "loss": 1.183,
+      "step": 1665
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.898082744702321e-05,
+      "loss": 1.2052,
+      "step": 1670
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.9157416750756805e-05,
+      "loss": 1.1974,
+      "step": 1675
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.933400605449041e-05,
+      "loss": 1.2046,
+      "step": 1680
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.951059535822401e-05,
+      "loss": 1.2191,
+      "step": 1685
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.968718466195761e-05,
+      "loss": 1.1759,
+      "step": 1690
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.986377396569122e-05,
+      "loss": 1.1929,
+      "step": 1695
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.004036326942482e-05,
+      "loss": 1.2096,
+      "step": 1700
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.021695257315842e-05,
+      "loss": 1.1843,
+      "step": 1705
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.039354187689202e-05,
+      "loss": 1.22,
+      "step": 1710
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.057013118062563e-05,
+      "loss": 1.2437,
+      "step": 1715
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.074672048435923e-05,
+      "loss": 1.2405,
+      "step": 1720
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.092330978809283e-05,
+      "loss": 1.1977,
+      "step": 1725
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.109989909182643e-05,
+      "loss": 1.2082,
+      "step": 1730
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.127648839556002e-05,
+      "loss": 1.1768,
+      "step": 1735
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.145307769929364e-05,
+      "loss": 1.2106,
+      "step": 1740
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.162966700302723e-05,
+      "loss": 1.2079,
+      "step": 1745
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.180625630676085e-05,
+      "loss": 1.2055,
+      "step": 1750
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 6.198284561049445e-05,
+      "loss": 1.1535,
+      "step": 1755
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 6.215943491422804e-05,
+      "loss": 1.1937,
+      "step": 1760
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 6.233602421796166e-05,
+      "loss": 1.1952,
+      "step": 1765
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 6.251261352169525e-05,
+      "loss": 1.1942,
+      "step": 1770
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 6.268920282542885e-05,
+      "loss": 1.1944,
+      "step": 1775
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 6.286579212916245e-05,
+      "loss": 1.2282,
+      "step": 1780
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 6.304238143289606e-05,
+      "loss": 1.2229,
+      "step": 1785
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 6.321897073662966e-05,
+      "loss": 1.2165,
+      "step": 1790
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 6.339556004036326e-05,
+      "loss": 1.1612,
+      "step": 1795
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 6.357214934409687e-05,
+      "loss": 1.1903,
+      "step": 1800
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 6.374873864783047e-05,
+      "loss": 1.2134,
+      "step": 1805
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 6.392532795156408e-05,
+      "loss": 1.1939,
+      "step": 1810
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 6.410191725529768e-05,
+      "loss": 1.1374,
+      "step": 1815
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.427850655903128e-05,
+      "loss": 1.2071,
+      "step": 1820
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.445509586276488e-05,
+      "loss": 1.1706,
+      "step": 1825
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.463168516649847e-05,
+      "loss": 1.1607,
+      "step": 1830
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.480827447023209e-05,
+      "loss": 1.1973,
+      "step": 1835
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.498486377396568e-05,
+      "loss": 1.2105,
+      "step": 1840
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.516145307769928e-05,
+      "loss": 1.2281,
+      "step": 1845
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.53380423814329e-05,
+      "loss": 1.1756,
+      "step": 1850
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.551463168516649e-05,
+      "loss": 1.1762,
+      "step": 1855
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.56912209889001e-05,
+      "loss": 1.2148,
+      "step": 1860
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.58678102926337e-05,
+      "loss": 1.2064,
+      "step": 1865
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.60443995963673e-05,
+      "loss": 1.1657,
+      "step": 1870
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.62209889001009e-05,
+      "loss": 1.2323,
+      "step": 1875
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.63975782038345e-05,
+      "loss": 1.1943,
+      "step": 1880
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.657416750756811e-05,
+      "loss": 1.1784,
+      "step": 1885
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.675075681130171e-05,
+      "loss": 1.2004,
+      "step": 1890
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.69273461150353e-05,
+      "loss": 1.2294,
+      "step": 1895
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.710393541876892e-05,
+      "loss": 1.1536,
+      "step": 1900
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.728052472250252e-05,
+      "loss": 1.1974,
+      "step": 1905
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.745711402623613e-05,
+      "loss": 1.1578,
+      "step": 1910
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.763370332996973e-05,
+      "loss": 1.1783,
+      "step": 1915
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.781029263370332e-05,
+      "loss": 1.1791,
+      "step": 1920
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.798688193743692e-05,
+      "loss": 1.1734,
+      "step": 1925
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.816347124117052e-05,
+      "loss": 1.1811,
+      "step": 1930
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.834006054490413e-05,
+      "loss": 1.1807,
+      "step": 1935
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.851664984863773e-05,
+      "loss": 1.1689,
+      "step": 1940
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.869323915237134e-05,
+      "loss": 1.185,
+      "step": 1945
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.886982845610494e-05,
+      "loss": 1.1644,
+      "step": 1950
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.904641775983854e-05,
+      "loss": 1.1928,
+      "step": 1955
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.922300706357215e-05,
+      "loss": 1.1911,
+      "step": 1960
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.939959636730575e-05,
+      "loss": 1.1903,
+      "step": 1965
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.957618567103935e-05,
+      "loss": 1.1952,
+      "step": 1970
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.975277497477295e-05,
+      "loss": 1.2188,
+      "step": 1975
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.992936427850656e-05,
+      "loss": 1.2063,
+      "step": 1980
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.999999511364106e-05,
+      "loss": 1.1944,
+      "step": 1985
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.999996525256358e-05,
+      "loss": 1.146,
+      "step": 1990
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.999990824507563e-05,
+      "loss": 1.22,
+      "step": 1995
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.99998240912214e-05,
+      "loss": 1.1721,
+      "step": 2000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.999971279106618e-05,
+      "loss": 1.1641,
+      "step": 2005
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.999957434469628e-05,
+      "loss": 1.1521,
+      "step": 2010
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.99994087522191e-05,
+      "loss": 1.1637,
+      "step": 2015
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.999921601376306e-05,
+      "loss": 1.1661,
+      "step": 2020
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.999899612947765e-05,
+      "loss": 1.1747,
+      "step": 2025
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.999874909953342e-05,
+      "loss": 1.1429,
+      "step": 2030
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.999847492412199e-05,
+      "loss": 1.2097,
+      "step": 2035
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.999817360345597e-05,
+      "loss": 1.198,
+      "step": 2040
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.99978451377691e-05,
+      "loss": 1.2063,
+      "step": 2045
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.999748952731613e-05,
+      "loss": 1.17,
+      "step": 2050
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.999710677237287e-05,
+      "loss": 1.1945,
+      "step": 2055
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.999669687323621e-05,
+      "loss": 1.1826,
+      "step": 2060
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.999625983022407e-05,
+      "loss": 1.2039,
+      "step": 2065
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.999579564367541e-05,
+      "loss": 1.1694,
+      "step": 2070
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.999530431395027e-05,
+      "loss": 1.1808,
+      "step": 2075
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.999478584142972e-05,
+      "loss": 1.2009,
+      "step": 2080
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.99942402265159e-05,
+      "loss": 1.172,
+      "step": 2085
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.999366746963202e-05,
+      "loss": 1.15,
+      "step": 2090
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.999306757122228e-05,
+      "loss": 1.2181,
+      "step": 2095
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.999244053175198e-05,
+      "loss": 1.1597,
+      "step": 2100
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.999178635170746e-05,
+      "loss": 1.1499,
+      "step": 2105
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.999110503159612e-05,
+      "loss": 1.1578,
+      "step": 2110
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.99903965719464e-05,
+      "loss": 1.1567,
+      "step": 2115
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.998966097330778e-05,
+      "loss": 1.1534,
+      "step": 2120
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.998889823625081e-05,
+      "loss": 1.1666,
+      "step": 2125
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.998810836136705e-05,
+      "loss": 1.179,
+      "step": 2130
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.998729134926918e-05,
+      "loss": 1.1818,
+      "step": 2135
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.998644720059085e-05,
+      "loss": 1.1437,
+      "step": 2140
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.998557591598682e-05,
+      "loss": 1.1761,
+      "step": 2145
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.998467749613284e-05,
+      "loss": 1.2009,
+      "step": 2150
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.998375194172577e-05,
+      "loss": 1.1979,
+      "step": 2155
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.998279925348345e-05,
+      "loss": 1.153,
+      "step": 2160
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.998181943214481e-05,
+      "loss": 1.1707,
+      "step": 2165
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.998081247846982e-05,
+      "loss": 1.1574,
+      "step": 2170
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.997977839323946e-05,
+      "loss": 1.1209,
+      "step": 2175
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.997871717725581e-05,
+      "loss": 1.1912,
+      "step": 2180
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.997762883134195e-05,
+      "loss": 1.1508,
+      "step": 2185
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.997651335634201e-05,
+      "loss": 1.1797,
+      "step": 2190
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.997537075312116e-05,
+      "loss": 1.1761,
+      "step": 2195
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.997420102256564e-05,
+      "loss": 1.1851,
+      "step": 2200
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.997300416558268e-05,
+      "loss": 1.1652,
+      "step": 2205
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.99717801831006e-05,
+      "loss": 1.1339,
+      "step": 2210
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.997052907606873e-05,
+      "loss": 1.1245,
+      "step": 2215
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.996925084545745e-05,
+      "loss": 1.1951,
+      "step": 2220
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.996794549225815e-05,
+      "loss": 1.1473,
+      "step": 2225
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.99666130174833e-05,
+      "loss": 1.1264,
+      "step": 2230
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.996525342216637e-05,
+      "loss": 1.1519,
+      "step": 2235
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.996386670736189e-05,
+      "loss": 1.1645,
+      "step": 2240
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.99624528741454e-05,
+      "loss": 1.1108,
+      "step": 2245
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.996101192361351e-05,
+      "loss": 1.1917,
+      "step": 2250
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.995954385688382e-05,
+      "loss": 1.1409,
+      "step": 2255
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.995804867509498e-05,
+      "loss": 1.1738,
+      "step": 2260
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.995652637940669e-05,
+      "loss": 1.1848,
+      "step": 2265
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.995497697099964e-05,
+      "loss": 1.1533,
+      "step": 2270
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.995340045107558e-05,
+      "loss": 1.1622,
+      "step": 2275
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.995179682085727e-05,
+      "loss": 1.1583,
+      "step": 2280
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.995016608158852e-05,
+      "loss": 1.152,
+      "step": 2285
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.994850823453416e-05,
+      "loss": 1.1577,
+      "step": 2290
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.994682328098e-05,
+      "loss": 1.1535,
+      "step": 2295
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.994511122223296e-05,
+      "loss": 1.1199,
+      "step": 2300
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.994337205962087e-05,
+      "loss": 1.1482,
+      "step": 2305
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.994160579449271e-05,
+      "loss": 1.1664,
+      "step": 2310
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.993981242821838e-05,
+      "loss": 1.1513,
+      "step": 2315
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.993799196218885e-05,
+      "loss": 1.1433,
+      "step": 2320
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.993614439781611e-05,
+      "loss": 1.1352,
+      "step": 2325
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.993426973653313e-05,
+      "loss": 1.1478,
+      "step": 2330
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.993236797979391e-05,
+      "loss": 1.1873,
+      "step": 2335
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.993043912907352e-05,
+      "loss": 1.161,
+      "step": 2340
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.992848318586797e-05,
+      "loss": 1.1252,
+      "step": 2345
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.99265001516943e-05,
+      "loss": 1.1408,
+      "step": 2350
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.992449002809063e-05,
+      "loss": 1.1594,
+      "step": 2355
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.992245281661598e-05,
+      "loss": 1.1673,
+      "step": 2360
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.992038851885047e-05,
+      "loss": 1.1661,
+      "step": 2365
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.991829713639518e-05,
+      "loss": 1.1705,
+      "step": 2370
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.991617867087222e-05,
+      "loss": 1.1565,
+      "step": 2375
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.99140331239247e-05,
+      "loss": 1.1496,
+      "step": 2380
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.991186049721672e-05,
+      "loss": 1.1837,
+      "step": 2385
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.990966079243341e-05,
+      "loss": 1.1326,
+      "step": 2390
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.990743401128089e-05,
+      "loss": 1.1382,
+      "step": 2395
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.990518015548626e-05,
+      "loss": 1.1627,
+      "step": 2400
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.990289922679766e-05,
+      "loss": 1.1686,
+      "step": 2405
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.990059122698419e-05,
+      "loss": 1.1332,
+      "step": 2410
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.989825615783599e-05,
+      "loss": 1.1577,
+      "step": 2415
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.989589402116414e-05,
+      "loss": 1.1636,
+      "step": 2420
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.989350481880075e-05,
+      "loss": 1.145,
+      "step": 2425
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.989108855259893e-05,
+      "loss": 1.1493,
+      "step": 2430
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.988864522443274e-05,
+      "loss": 1.1522,
+      "step": 2435
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.988617483619728e-05,
+      "loss": 1.1797,
+      "step": 2440
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.988367738980861e-05,
+      "loss": 1.1432,
+      "step": 2445
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.988115288720377e-05,
+      "loss": 1.1521,
+      "step": 2450
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.98786013303408e-05,
+      "loss": 1.1835,
+      "step": 2455
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.987602272119872e-05,
+      "loss": 1.1889,
+      "step": 2460
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.987341706177753e-05,
+      "loss": 1.156,
+      "step": 2465
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.987078435409823e-05,
+      "loss": 1.148,
+      "step": 2470
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.986812460020274e-05,
+      "loss": 1.1121,
+      "step": 2475
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.986543780215404e-05,
+      "loss": 1.1472,
+      "step": 2480
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.986272396203603e-05,
+      "loss": 1.1755,
+      "step": 2485
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.98599830819536e-05,
+      "loss": 1.1067,
+      "step": 2490
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.985721516403259e-05,
+      "loss": 1.1384,
+      "step": 2495
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.985442021041986e-05,
+      "loss": 1.154,
+      "step": 2500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.98515982232832e-05,
+      "loss": 1.1341,
+      "step": 2505
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.984874920481138e-05,
+      "loss": 1.1521,
+      "step": 2510
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.984587315721413e-05,
+      "loss": 1.1249,
+      "step": 2515
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.984297008272215e-05,
+      "loss": 1.1046,
+      "step": 2520
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.984003998358709e-05,
+      "loss": 1.1508,
+      "step": 2525
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.98370828620816e-05,
+      "loss": 1.1092,
+      "step": 2530
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.983409872049925e-05,
+      "loss": 1.1597,
+      "step": 2535
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.983108756115455e-05,
+      "loss": 1.145,
+      "step": 2540
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.982804938638301e-05,
+      "loss": 1.1416,
+      "step": 2545
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.982498419854111e-05,
+      "loss": 1.1412,
+      "step": 2550
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.98218920000062e-05,
+      "loss": 1.1443,
+      "step": 2555
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.981877279317665e-05,
+      "loss": 1.1489,
+      "step": 2560
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.981562658047175e-05,
+      "loss": 1.1216,
+      "step": 2565
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.981245336433174e-05,
+      "loss": 1.0985,
+      "step": 2570
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.980925314721781e-05,
+      "loss": 1.1038,
+      "step": 2575
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.98060259316121e-05,
+      "loss": 1.1404,
+      "step": 2580
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.980277172001766e-05,
+      "loss": 1.1419,
+      "step": 2585
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.97994905149585e-05,
+      "loss": 1.098,
+      "step": 2590
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.979618231897957e-05,
+      "loss": 1.113,
+      "step": 2595
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.979284713464674e-05,
+      "loss": 1.1281,
+      "step": 2600
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.978948496454685e-05,
+      "loss": 1.1013,
+      "step": 2605
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.97860958112876e-05,
+      "loss": 1.1614,
+      "step": 2610
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.978267967749768e-05,
+      "loss": 1.1305,
+      "step": 2615
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.977923656582669e-05,
+      "loss": 1.144,
+      "step": 2620
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.977576647894513e-05,
+      "loss": 1.1523,
+      "step": 2625
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.977226941954446e-05,
+      "loss": 1.1334,
+      "step": 2630
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.976874539033705e-05,
+      "loss": 1.0973,
+      "step": 2635
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.976519439405617e-05,
+      "loss": 1.1353,
+      "step": 2640
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.976161643345601e-05,
+      "loss": 1.1152,
+      "step": 2645
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.97580115113117e-05,
+      "loss": 1.1501,
+      "step": 2650
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.975437963041923e-05,
+      "loss": 1.1262,
+      "step": 2655
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.975072079359557e-05,
+      "loss": 1.1547,
+      "step": 2660
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.974703500367855e-05,
+      "loss": 1.1012,
+      "step": 2665
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.97433222635269e-05,
+      "loss": 1.1347,
+      "step": 2670
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.973958257602027e-05,
+      "loss": 1.1182,
+      "step": 2675
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.973581594405922e-05,
+      "loss": 1.1091,
+      "step": 2680
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.97320223705652e-05,
+      "loss": 1.1337,
+      "step": 2685
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.972820185848053e-05,
+      "loss": 1.108,
+      "step": 2690
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.972435441076846e-05,
+      "loss": 1.1259,
+      "step": 2695
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.972048003041313e-05,
+      "loss": 1.1144,
+      "step": 2700
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.971657872041955e-05,
+      "loss": 1.1334,
+      "step": 2705
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.971265048381361e-05,
+      "loss": 1.1021,
+      "step": 2710
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.97086953236421e-05,
+      "loss": 1.0966,
+      "step": 2715
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.970471324297272e-05,
+      "loss": 1.1094,
+      "step": 2720
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.970070424489402e-05,
+      "loss": 1.1244,
+      "step": 2725
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.969666833251539e-05,
+      "loss": 1.0941,
+      "step": 2730
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.969260550896717e-05,
+      "loss": 1.0975,
+      "step": 2735
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.968851577740053e-05,
+      "loss": 1.1436,
+      "step": 2740
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.968439914098749e-05,
+      "loss": 1.0992,
+      "step": 2745
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.9680255602921e-05,
+      "loss": 1.0909,
+      "step": 2750
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.967608516641482e-05,
+      "loss": 1.1153,
+      "step": 2755
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.96718878347036e-05,
+      "loss": 1.1195,
+      "step": 2760
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.966766361104285e-05,
+      "loss": 1.1516,
+      "step": 2765
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.966341249870891e-05,
+      "loss": 1.1373,
+      "step": 2770
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.965913450099902e-05,
+      "loss": 1.1491,
+      "step": 2775
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.965482962123123e-05,
+      "loss": 1.1403,
+      "step": 2780
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.965049786274446e-05,
+      "loss": 1.1218,
+      "step": 2785
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.964613922889849e-05,
+      "loss": 1.0917,
+      "step": 2790
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.964175372307392e-05,
+      "loss": 1.0972,
+      "step": 2795
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.96373413486722e-05,
+      "loss": 1.1132,
+      "step": 2800
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.963290210911564e-05,
+      "loss": 1.1119,
+      "step": 2805
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.962843600784734e-05,
+      "loss": 1.1192,
+      "step": 2810
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.96239430483313e-05,
+      "loss": 1.1138,
+      "step": 2815
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.961942323405228e-05,
+      "loss": 1.1269,
+      "step": 2820
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.961487656851593e-05,
+      "loss": 1.1085,
+      "step": 2825
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.96103030552487e-05,
+      "loss": 1.1312,
+      "step": 2830
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.960570269779785e-05,
+      "loss": 1.1549,
+      "step": 2835
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.960107549973148e-05,
+      "loss": 1.1141,
+      "step": 2840
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.95964214646385e-05,
+      "loss": 1.0991,
+      "step": 2845
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.959174059612864e-05,
+      "loss": 1.0915,
+      "step": 2850
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.958703289783246e-05,
+      "loss": 1.1113,
+      "step": 2855
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.958229837340126e-05,
+      "loss": 1.1083,
+      "step": 2860
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.957753702650725e-05,
+      "loss": 1.1168,
+      "step": 2865
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.957274886084336e-05,
+      "loss": 1.0966,
+      "step": 2870
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.956793388012336e-05,
+      "loss": 1.1247,
+      "step": 2875
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.956309208808181e-05,
+      "loss": 1.1121,
+      "step": 2880
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.955822348847407e-05,
+      "loss": 1.0917,
+      "step": 2885
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.955332808507627e-05,
+      "loss": 1.1263,
+      "step": 2890
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.954840588168535e-05,
+      "loss": 1.0999,
+      "step": 2895
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.954345688211905e-05,
+      "loss": 1.0928,
+      "step": 2900
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.953848109021585e-05,
+      "loss": 1.0978,
+      "step": 2905
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.953347850983507e-05,
+      "loss": 1.1267,
+      "step": 2910
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.952844914485674e-05,
+      "loss": 1.1202,
+      "step": 2915
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.95233929991817e-05,
+      "loss": 1.1237,
+      "step": 2920
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.951831007673159e-05,
+      "loss": 1.099,
+      "step": 2925
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.951320038144876e-05,
+      "loss": 1.1298,
+      "step": 2930
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.950806391729638e-05,
+      "loss": 1.1062,
+      "step": 2935
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.950290068825832e-05,
+      "loss": 1.1105,
+      "step": 2940
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.949771069833927e-05,
+      "loss": 1.1181,
+      "step": 2945
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.949249395156465e-05,
+      "loss": 1.0981,
+      "step": 2950
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.948725045198063e-05,
+      "loss": 1.0937,
+      "step": 2955
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.948198020365413e-05,
+      "loss": 1.1333,
+      "step": 2960
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.947668321067282e-05,
+      "loss": 1.1178,
+      "step": 2965
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.947135947714513e-05,
+      "loss": 1.0837,
+      "step": 2970
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.946600900720018e-05,
+      "loss": 1.1078,
+      "step": 2975
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.94606318049879e-05,
+      "loss": 1.1148,
+      "step": 2980
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.94552278746789e-05,
+      "loss": 1.1454,
+      "step": 2985
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.944979722046453e-05,
+      "loss": 1.087,
+      "step": 2990
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.944433984655687e-05,
+      "loss": 1.0992,
+      "step": 2995
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.943885575718875e-05,
+      "loss": 1.0767,
+      "step": 3000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.943334495661369e-05,
+      "loss": 1.1205,
+      "step": 3005
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.942780744910591e-05,
+      "loss": 1.0893,
+      "step": 3010
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.942224323896039e-05,
+      "loss": 1.139,
+      "step": 3015
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.94166523304928e-05,
+      "loss": 1.092,
+      "step": 3020
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.941103472803953e-05,
+      "loss": 1.1201,
+      "step": 3025
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.940539043595764e-05,
+      "loss": 1.1176,
+      "step": 3030
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.939971945862492e-05,
+      "loss": 1.079,
+      "step": 3035
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.939402180043988e-05,
+      "loss": 1.0814,
+      "step": 3040
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.938829746582166e-05,
+      "loss": 1.1013,
+      "step": 3045
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.938254645921012e-05,
+      "loss": 1.1034,
+      "step": 3050
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.937676878506585e-05,
+      "loss": 1.0905,
+      "step": 3055
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.937096444787006e-05,
+      "loss": 1.1084,
+      "step": 3060
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.936513345212466e-05,
+      "loss": 1.0932,
+      "step": 3065
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.935927580235227e-05,
+      "loss": 1.081,
+      "step": 3070
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.935339150309613e-05,
+      "loss": 1.1239,
+      "step": 3075
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.934748055892018e-05,
+      "loss": 1.1352,
+      "step": 3080
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.934154297440905e-05,
+      "loss": 1.1001,
+      "step": 3085
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.933557875416796e-05,
+      "loss": 1.086,
+      "step": 3090
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.932958790282287e-05,
+      "loss": 1.1047,
+      "step": 3095
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.932357042502036e-05,
+      "loss": 1.1074,
+      "step": 3100
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.931752632542763e-05,
+      "loss": 1.0966,
+      "step": 3105
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.931145560873259e-05,
+      "loss": 1.1089,
+      "step": 3110
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.930535827964373e-05,
+      "loss": 1.1214,
+      "step": 3115
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.929923434289026e-05,
+      "loss": 1.0784,
+      "step": 3120
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.929308380322193e-05,
+      "loss": 1.1481,
+      "step": 3125
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.928690666540923e-05,
+      "loss": 1.1025,
+      "step": 3130
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.928070293424318e-05,
+      "loss": 1.1231,
+      "step": 3135
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.927447261453551e-05,
+      "loss": 1.0755,
+      "step": 3140
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.926821571111848e-05,
+      "loss": 1.1274,
+      "step": 3145
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.926193222884509e-05,
+      "loss": 1.1154,
+      "step": 3150
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.925562217258883e-05,
+      "loss": 1.1601,
+      "step": 3155
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.924928554724389e-05,
+      "loss": 1.1175,
+      "step": 3160
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.924292235772503e-05,
+      "loss": 1.1203,
+      "step": 3165
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.923653260896762e-05,
+      "loss": 1.0944,
+      "step": 3170
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.923011630592762e-05,
+      "loss": 1.0701,
+      "step": 3175
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.92236734535816e-05,
+      "loss": 1.0875,
+      "step": 3180
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.921720405692671e-05,
+      "loss": 1.0809,
+      "step": 3185
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.921070812098071e-05,
+      "loss": 1.0885,
+      "step": 3190
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.920418565078192e-05,
+      "loss": 1.1334,
+      "step": 3195
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.919763665138925e-05,
+      "loss": 1.1283,
+      "step": 3200
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.91910611278822e-05,
+      "loss": 1.1239,
+      "step": 3205
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.91844590853608e-05,
+      "loss": 1.091,
+      "step": 3210
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.91778305289457e-05,
+      "loss": 1.0899,
+      "step": 3215
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.917117546377807e-05,
+      "loss": 1.0397,
+      "step": 3220
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.91644938950197e-05,
+      "loss": 1.1092,
+      "step": 3225
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.915778582785284e-05,
+      "loss": 1.1141,
+      "step": 3230
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.91510512674804e-05,
+      "loss": 1.1289,
+      "step": 3235
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.914429021912578e-05,
+      "loss": 1.0883,
+      "step": 3240
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.913750268803292e-05,
+      "loss": 1.0828,
+      "step": 3245
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.91306886794663e-05,
+      "loss": 1.073,
+      "step": 3250
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.9123848198711e-05,
+      "loss": 1.0899,
+      "step": 3255
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.911698125107255e-05,
+      "loss": 1.0555,
+      "step": 3260
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.911008784187702e-05,
+      "loss": 1.0989,
+      "step": 3265
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 6.910316797647108e-05,
+      "loss": 1.0465,
+      "step": 3270
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 6.90962216602218e-05,
+      "loss": 1.0889,
+      "step": 3275
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 6.908924889851687e-05,
+      "loss": 1.0883,
+      "step": 3280
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 6.908224969676445e-05,
+      "loss": 1.0863,
+      "step": 3285
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 6.90752240603932e-05,
+      "loss": 1.1088,
+      "step": 3290
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 6.906817199485229e-05,
+      "loss": 1.0928,
+      "step": 3295
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 6.906109350561138e-05,
+      "loss": 1.0499,
+      "step": 3300
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 6.905398859816067e-05,
+      "loss": 1.0996,
+      "step": 3305
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 6.904685727801079e-05,
+      "loss": 1.0806,
+      "step": 3310
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 6.903969955069287e-05,
+      "loss": 1.086,
+      "step": 3315
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 6.903251542175855e-05,
+      "loss": 1.1084,
+      "step": 3320
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 6.902530489677991e-05,
+      "loss": 1.1111,
+      "step": 3325
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 6.901806798134958e-05,
+      "loss": 1.1213,
+      "step": 3330
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 6.901080468108052e-05,
+      "loss": 1.0715,
+      "step": 3335
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 6.90035150016063e-05,
+      "loss": 1.1262,
+      "step": 3340
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 6.899619894858086e-05,
+      "loss": 1.0876,
+      "step": 3345
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 6.898885652767863e-05,
+      "loss": 1.0658,
+      "step": 3350
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 6.898148774459448e-05,
+      "loss": 1.0946,
+      "step": 3355
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 6.897409260504372e-05,
+      "loss": 1.0779,
+      "step": 3360
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 6.896667111476213e-05,
+      "loss": 1.0611,
+      "step": 3365
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 6.895922327950589e-05,
+      "loss": 1.1128,
+      "step": 3370
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 6.895174910505167e-05,
+      "loss": 1.1214,
+      "step": 3375
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 6.89442485971965e-05,
+      "loss": 1.0846,
+      "step": 3380
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 6.893672176175788e-05,
+      "loss": 1.0483,
+      "step": 3385
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 6.89291686045737e-05,
+      "loss": 1.101,
+      "step": 3390
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 6.892158913150233e-05,
+      "loss": 1.0794,
+      "step": 3395
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 6.891398334842245e-05,
+      "loss": 1.1159,
+      "step": 3400
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 6.890635126123324e-05,
+      "loss": 1.0801,
+      "step": 3405
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 6.88986928758542e-05,
+      "loss": 1.0609,
+      "step": 3410
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 6.889100819822532e-05,
+      "loss": 1.1042,
+      "step": 3415
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 6.888329723430691e-05,
+      "loss": 1.1107,
+      "step": 3420
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 6.887555999007966e-05,
+      "loss": 1.0861,
+      "step": 3425
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 6.886779647154473e-05,
+      "loss": 1.0599,
+      "step": 3430
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 6.886000668472356e-05,
+      "loss": 1.0738,
+      "step": 3435
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 6.885219063565801e-05,
+      "loss": 1.0714,
+      "step": 3440
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 6.884434833041031e-05,
+      "loss": 1.1052,
+      "step": 3445
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 6.883647977506308e-05,
+      "loss": 1.0607,
+      "step": 3450
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 6.882858497571922e-05,
+      "loss": 1.068,
+      "step": 3455
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 6.882066393850206e-05,
+      "loss": 1.0403,
+      "step": 3460
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 6.881271666955528e-05,
+      "loss": 1.0754,
+      "step": 3465
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.880474317504283e-05,
+      "loss": 1.0628,
+      "step": 3470
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.879674346114909e-05,
+      "loss": 1.0633,
+      "step": 3475
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.878871753407871e-05,
+      "loss": 1.098,
+      "step": 3480
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.878066540005673e-05,
+      "loss": 1.0993,
+      "step": 3485
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.877258706532846e-05,
+      "loss": 1.0991,
+      "step": 3490
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.876448253615958e-05,
+      "loss": 1.1018,
+      "step": 3495
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.875635181883604e-05,
+      "loss": 1.0831,
+      "step": 3500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.874819491966415e-05,
+      "loss": 1.0919,
+      "step": 3505
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.874001184497048e-05,
+      "loss": 1.0514,
+      "step": 3510
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.873180260110194e-05,
+      "loss": 1.0827,
+      "step": 3515
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.872356719442572e-05,
+      "loss": 1.0548,
+      "step": 3520
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.87153056313293e-05,
+      "loss": 1.1282,
+      "step": 3525
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.870701791822045e-05,
+      "loss": 1.1026,
+      "step": 3530
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.869870406152723e-05,
+      "loss": 1.0966,
+      "step": 3535
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.869036406769797e-05,
+      "loss": 1.0723,
+      "step": 3540
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.868199794320128e-05,
+      "loss": 1.0434,
+      "step": 3545
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.867360569452601e-05,
+      "loss": 1.0371,
+      "step": 3550
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.866518732818132e-05,
+      "loss": 1.0915,
+      "step": 3555
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.865674285069658e-05,
+      "loss": 1.1227,
+      "step": 3560
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.864827226862142e-05,
+      "loss": 1.0952,
+      "step": 3565
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.863977558852578e-05,
+      "loss": 1.0634,
+      "step": 3570
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.863125281699974e-05,
+      "loss": 1.0626,
+      "step": 3575
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.86227039606537e-05,
+      "loss": 1.0927,
+      "step": 3580
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.861412902611823e-05,
+      "loss": 1.0778,
+      "step": 3585
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.860552802004419e-05,
+      "loss": 1.0752,
+      "step": 3590
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.85969009491026e-05,
+      "loss": 1.0785,
+      "step": 3595
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.858824781998474e-05,
+      "loss": 1.0879,
+      "step": 3600
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.857956863940208e-05,
+      "loss": 1.0846,
+      "step": 3605
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.857086341408631e-05,
+      "loss": 1.0546,
+      "step": 3610
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.856213215078929e-05,
+      "loss": 1.041,
+      "step": 3615
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.855337485628312e-05,
+      "loss": 1.0403,
+      "step": 3620
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.854459153736006e-05,
+      "loss": 1.0766,
+      "step": 3625
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.853578220083254e-05,
+      "loss": 1.106,
+      "step": 3630
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.852694685353323e-05,
+      "loss": 1.0696,
+      "step": 3635
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.85180855023149e-05,
+      "loss": 1.0865,
+      "step": 3640
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.850919815405055e-05,
+      "loss": 1.1023,
+      "step": 3645
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.85002848156333e-05,
+      "loss": 1.0889,
+      "step": 3650
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.849134549397646e-05,
+      "loss": 1.0883,
+      "step": 3655
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.848238019601349e-05,
+      "loss": 1.0857,
+      "step": 3660
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.847338892869796e-05,
+      "loss": 1.0952,
+      "step": 3665
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.846437169900362e-05,
+      "loss": 1.0403,
+      "step": 3670
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.845532851392435e-05,
+      "loss": 1.1051,
+      "step": 3675
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.844625938047416e-05,
+      "loss": 1.0738,
+      "step": 3680
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.843716430568717e-05,
+      "loss": 1.0303,
+      "step": 3685
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.842804329661766e-05,
+      "loss": 1.1036,
+      "step": 3690
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.841889636033996e-05,
+      "loss": 1.0894,
+      "step": 3695
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.84097235039486e-05,
+      "loss": 1.0517,
+      "step": 3700
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.84005247345581e-05,
+      "loss": 1.0739,
+      "step": 3705
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.839130005930319e-05,
+      "loss": 1.0653,
+      "step": 3710
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.838204948533861e-05,
+      "loss": 1.0987,
+      "step": 3715
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.837277301983924e-05,
+      "loss": 1.0865,
+      "step": 3720
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.836347067000001e-05,
+      "loss": 1.0427,
+      "step": 3725
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.835414244303597e-05,
+      "loss": 1.0609,
+      "step": 3730
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.834478834618218e-05,
+      "loss": 1.0813,
+      "step": 3735
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.83354083866938e-05,
+      "loss": 1.0508,
+      "step": 3740
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.832600257184604e-05,
+      "loss": 1.1095,
+      "step": 3745
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.831657090893417e-05,
+      "loss": 1.0713,
+      "step": 3750
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.83071134052735e-05,
+      "loss": 1.0929,
+      "step": 3755
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.829763006819942e-05,
+      "loss": 1.0592,
+      "step": 3760
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.828812090506728e-05,
+      "loss": 1.0739,
+      "step": 3765
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.827858592325254e-05,
+      "loss": 1.0498,
+      "step": 3770
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.826902513015064e-05,
+      "loss": 1.0597,
+      "step": 3775
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.825943853317704e-05,
+      "loss": 1.0419,
+      "step": 3780
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.824982613976724e-05,
+      "loss": 1.0743,
+      "step": 3785
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.824018795737672e-05,
+      "loss": 1.0756,
+      "step": 3790
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.823052399348099e-05,
+      "loss": 1.1052,
+      "step": 3795
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.82208342555755e-05,
+      "loss": 1.0492,
+      "step": 3800
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.82111187511758e-05,
+      "loss": 1.0519,
+      "step": 3805
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.820137748781731e-05,
+      "loss": 1.0557,
+      "step": 3810
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.819161047305547e-05,
+      "loss": 1.0849,
+      "step": 3815
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.818181771446572e-05,
+      "loss": 1.0522,
+      "step": 3820
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.817199921964345e-05,
+      "loss": 1.0737,
+      "step": 3825
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.816215499620398e-05,
+      "loss": 1.0823,
+      "step": 3830
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.815228505178265e-05,
+      "loss": 1.059,
+      "step": 3835
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.814238939403469e-05,
+      "loss": 1.0491,
+      "step": 3840
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.813246803063529e-05,
+      "loss": 1.0262,
+      "step": 3845
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.812252096927958e-05,
+      "loss": 1.084,
+      "step": 3850
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.811254821768266e-05,
+      "loss": 1.0961,
+      "step": 3855
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.810254978357948e-05,
+      "loss": 1.0663,
+      "step": 3860
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.809252567472498e-05,
+      "loss": 1.0767,
+      "step": 3865
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.808247589889396e-05,
+      "loss": 1.1185,
+      "step": 3870
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.807240046388118e-05,
+      "loss": 1.0902,
+      "step": 3875
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.806229937750125e-05,
+      "loss": 1.0645,
+      "step": 3880
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.80521726475887e-05,
+      "loss": 1.0407,
+      "step": 3885
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.804202028199795e-05,
+      "loss": 1.0356,
+      "step": 3890
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.803184228860333e-05,
+      "loss": 1.0741,
+      "step": 3895
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.802163867529896e-05,
+      "loss": 1.0574,
+      "step": 3900
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.801140944999895e-05,
+      "loss": 1.068,
+      "step": 3905
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.800115462063719e-05,
+      "loss": 1.0736,
+      "step": 3910
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.799087419516743e-05,
+      "loss": 1.0492,
+      "step": 3915
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.798056818156335e-05,
+      "loss": 1.0424,
+      "step": 3920
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.797023658781836e-05,
+      "loss": 1.0825,
+      "step": 3925
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.795987942194583e-05,
+      "loss": 1.0217,
+      "step": 3930
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.794949669197888e-05,
+      "loss": 1.0592,
+      "step": 3935
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.793908840597049e-05,
+      "loss": 1.0421,
+      "step": 3940
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.792865457199345e-05,
+      "loss": 1.0661,
+      "step": 3945
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.791819519814038e-05,
+      "loss": 1.0769,
+      "step": 3950
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.79077102925237e-05,
+      "loss": 1.0683,
+      "step": 3955
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.789719986327563e-05,
+      "loss": 1.0187,
+      "step": 3960
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.78866639185482e-05,
+      "loss": 1.0599,
+      "step": 3965
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.787610246651319e-05,
+      "loss": 1.0773,
+      "step": 3970
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.786551551536224e-05,
+      "loss": 1.0507,
+      "step": 3975
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.785490307330669e-05,
+      "loss": 1.0099,
+      "step": 3980
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.784426514857769e-05,
+      "loss": 1.0504,
+      "step": 3985
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.783360174942615e-05,
+      "loss": 1.064,
+      "step": 3990
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.782291288412273e-05,
+      "loss": 1.0744,
+      "step": 3995
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.781219856095787e-05,
+      "loss": 1.0936,
+      "step": 4000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.78014587882417e-05,
+      "loss": 1.0871,
+      "step": 4005
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.779069357430414e-05,
+      "loss": 1.0449,
+      "step": 4010
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.777990292749482e-05,
+      "loss": 1.0453,
+      "step": 4015
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.77690868561831e-05,
+      "loss": 1.0729,
+      "step": 4020
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.775824536875806e-05,
+      "loss": 1.1021,
+      "step": 4025
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.774737847362851e-05,
+      "loss": 1.0474,
+      "step": 4030
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.773648617922293e-05,
+      "loss": 1.0732,
+      "step": 4035
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.772556849398952e-05,
+      "loss": 1.0597,
+      "step": 4040
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.771462542639617e-05,
+      "loss": 1.0703,
+      "step": 4045
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.770365698493047e-05,
+      "loss": 1.0525,
+      "step": 4050
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.769266317809969e-05,
+      "loss": 1.069,
+      "step": 4055
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.768164401443075e-05,
+      "loss": 1.0465,
+      "step": 4060
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.767059950247023e-05,
+      "loss": 1.008,
+      "step": 4065
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.765952965078445e-05,
+      "loss": 1.046,
+      "step": 4070
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.764843446795926e-05,
+      "loss": 1.0632,
+      "step": 4075
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.763731396260027e-05,
+      "loss": 1.0452,
+      "step": 4080
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.762616814333265e-05,
+      "loss": 1.0339,
+      "step": 4085
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.761499701880127e-05,
+      "loss": 1.0459,
+      "step": 4090
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.760380059767057e-05,
+      "loss": 1.0875,
+      "step": 4095
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.759257888862463e-05,
+      "loss": 1.046,
+      "step": 4100
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.758133190036716e-05,
+      "loss": 1.0561,
+      "step": 4105
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.757005964162147e-05,
+      "loss": 1.0701,
+      "step": 4110
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.755876212113045e-05,
+      "loss": 1.0092,
+      "step": 4115
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.754743934765661e-05,
+      "loss": 1.0611,
+      "step": 4120
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.753609132998203e-05,
+      "loss": 1.0558,
+      "step": 4125
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.752471807690837e-05,
+      "loss": 1.0816,
+      "step": 4130
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.751331959725688e-05,
+      "loss": 1.0109,
+      "step": 4135
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.750189589986835e-05,
+      "loss": 1.0669,
+      "step": 4140
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.749044699360313e-05,
+      "loss": 1.0949,
+      "step": 4145
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.747897288734117e-05,
+      "loss": 1.0854,
+      "step": 4150
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.746747358998191e-05,
+      "loss": 1.051,
+      "step": 4155
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.745594911044435e-05,
+      "loss": 1.0578,
+      "step": 4160
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.744439945766703e-05,
+      "loss": 1.0338,
+      "step": 4165
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.743282464060798e-05,
+      "loss": 1.0384,
+      "step": 4170
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.74212246682448e-05,
+      "loss": 1.0613,
+      "step": 4175
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.740959954957456e-05,
+      "loss": 1.0743,
+      "step": 4180
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.739794929361386e-05,
+      "loss": 1.0153,
+      "step": 4185
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.738627390939875e-05,
+      "loss": 1.0305,
+      "step": 4190
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.737457340598485e-05,
+      "loss": 1.085,
+      "step": 4195
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.736284779244718e-05,
+      "loss": 1.0562,
+      "step": 4200
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.73510970778803e-05,
+      "loss": 1.0669,
+      "step": 4205
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.733932127139818e-05,
+      "loss": 1.0135,
+      "step": 4210
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.732752038213432e-05,
+      "loss": 1.1002,
+      "step": 4215
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.731569441924161e-05,
+      "loss": 1.0582,
+      "step": 4220
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.730384339189242e-05,
+      "loss": 1.0582,
+      "step": 4225
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.729196730927856e-05,
+      "loss": 1.0554,
+      "step": 4230
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.728006618061126e-05,
+      "loss": 1.0787,
+      "step": 4235
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.726814001512121e-05,
+      "loss": 1.0711,
+      "step": 4240
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.725618882205845e-05,
+      "loss": 1.0614,
+      "step": 4245
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.724421261069252e-05,
+      "loss": 1.0268,
+      "step": 4250
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.723221139031227e-05,
+      "loss": 1.0047,
+      "step": 4255
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.722018517022605e-05,
+      "loss": 1.0386,
+      "step": 4260
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.72081339597615e-05,
+      "loss": 1.0493,
+      "step": 4265
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.71960577682657e-05,
+      "loss": 1.0648,
+      "step": 4270
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.718395660510513e-05,
+      "loss": 1.0567,
+      "step": 4275
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.717183047966557e-05,
+      "loss": 1.0661,
+      "step": 4280
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.715967940135219e-05,
+      "loss": 1.038,
+      "step": 4285
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.714750337958954e-05,
+      "loss": 1.0179,
+      "step": 4290
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.713530242382149e-05,
+      "loss": 1.0677,
+      "step": 4295
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.712307654351125e-05,
+      "loss": 1.0268,
+      "step": 4300
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.711082574814136e-05,
+      "loss": 1.0375,
+      "step": 4305
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.709855004721368e-05,
+      "loss": 1.0734,
+      "step": 4310
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.708624945024943e-05,
+      "loss": 1.0317,
+      "step": 4315
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.707392396678909e-05,
+      "loss": 1.033,
+      "step": 4320
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.706157360639245e-05,
+      "loss": 1.0499,
+      "step": 4325
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.70491983786386e-05,
+      "loss": 1.0396,
+      "step": 4330
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.703679829312594e-05,
+      "loss": 1.0414,
+      "step": 4335
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.702437335947214e-05,
+      "loss": 1.0699,
+      "step": 4340
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.70119235873141e-05,
+      "loss": 1.0779,
+      "step": 4345
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.699944898630804e-05,
+      "loss": 1.0244,
+      "step": 4350
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.69869495661294e-05,
+      "loss": 1.0362,
+      "step": 4355
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.697442533647292e-05,
+      "loss": 1.0708,
+      "step": 4360
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.696187630705254e-05,
+      "loss": 1.0329,
+      "step": 4365
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.69493024876014e-05,
+      "loss": 1.0337,
+      "step": 4370
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.693670388787196e-05,
+      "loss": 1.0179,
+      "step": 4375
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.692408051763584e-05,
+      "loss": 1.0299,
+      "step": 4380
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.691143238668387e-05,
+      "loss": 1.0259,
+      "step": 4385
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.689875950482612e-05,
+      "loss": 1.0833,
+      "step": 4390
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.68860618818918e-05,
+      "loss": 1.0222,
+      "step": 4395
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.687333952772939e-05,
+      "loss": 1.0925,
+      "step": 4400
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.686059245220648e-05,
+      "loss": 1.0516,
+      "step": 4405
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.684782066520987e-05,
+      "loss": 1.0348,
+      "step": 4410
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.68350241766455e-05,
+      "loss": 1.0461,
+      "step": 4415
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.682220299643851e-05,
+      "loss": 1.0449,
+      "step": 4420
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.680935713453315e-05,
+      "loss": 1.0168,
+      "step": 4425
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.679648660089284e-05,
+      "loss": 1.03,
+      "step": 4430
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.678359140550013e-05,
+      "loss": 1.0277,
+      "step": 4435
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.677067155835671e-05,
+      "loss": 1.0674,
+      "step": 4440
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.675772706948333e-05,
+      "loss": 1.0602,
+      "step": 4445
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.674475794891992e-05,
+      "loss": 1.0378,
+      "step": 4450
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.67317642067255e-05,
+      "loss": 1.0007,
+      "step": 4455
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.671874585297818e-05,
+      "loss": 1.0671,
+      "step": 4460
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.670570289777515e-05,
+      "loss": 1.0667,
+      "step": 4465
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.669263535123269e-05,
+      "loss": 1.0345,
+      "step": 4470
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.667954322348615e-05,
+      "loss": 1.0581,
+      "step": 4475
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.666642652468994e-05,
+      "loss": 1.0124,
+      "step": 4480
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.665328526501757e-05,
+      "loss": 1.0247,
+      "step": 4485
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.664011945466152e-05,
+      "loss": 1.0749,
+      "step": 4490
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.66269291038334e-05,
+      "loss": 1.0291,
+      "step": 4495
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.661371422276377e-05,
+      "loss": 0.9877,
+      "step": 4500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.660047482170228e-05,
+      "loss": 1.067,
+      "step": 4505
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.658721091091759e-05,
+      "loss": 1.0353,
+      "step": 4510
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.657392250069732e-05,
+      "loss": 1.0248,
+      "step": 4515
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.656060960134815e-05,
+      "loss": 1.0532,
+      "step": 4520
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.654727222319571e-05,
+      "loss": 0.9973,
+      "step": 4525
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.653391037658466e-05,
+      "loss": 0.9974,
+      "step": 4530
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.652052407187862e-05,
+      "loss": 1.0209,
+      "step": 4535
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.650711331946014e-05,
+      "loss": 1.0511,
+      "step": 4540
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.64936781297308e-05,
+      "loss": 1.0552,
+      "step": 4545
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.64802185131111e-05,
+      "loss": 1.013,
+      "step": 4550
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.646673448004046e-05,
+      "loss": 1.0192,
+      "step": 4555
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.64532260409773e-05,
+      "loss": 1.0258,
+      "step": 4560
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.643969320639889e-05,
+      "loss": 1.0208,
+      "step": 4565
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.642613598680152e-05,
+      "loss": 1.0383,
+      "step": 4570
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.641255439270029e-05,
+      "loss": 1.0507,
+      "step": 4575
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.639894843462929e-05,
+      "loss": 1.0394,
+      "step": 4580
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.638531812314145e-05,
+      "loss": 1.003,
+      "step": 4585
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.637166346880861e-05,
+      "loss": 1.0229,
+      "step": 4590
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.635798448222151e-05,
+      "loss": 1.0394,
+      "step": 4595
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.634428117398972e-05,
+      "loss": 1.0283,
+      "step": 4600
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.63305535547417e-05,
+      "loss": 1.0434,
+      "step": 4605
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.631680163512478e-05,
+      "loss": 1.0447,
+      "step": 4610
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.63030254258051e-05,
+      "loss": 1.0395,
+      "step": 4615
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.628922493746769e-05,
+      "loss": 1.0133,
+      "step": 4620
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.627540018081635e-05,
+      "loss": 1.0278,
+      "step": 4625
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.626155116657375e-05,
+      "loss": 1.0125,
+      "step": 4630
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.624767790548135e-05,
+      "loss": 1.0406,
+      "step": 4635
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.623378040829941e-05,
+      "loss": 1.0306,
+      "step": 4640
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.621985868580704e-05,
+      "loss": 1.0667,
+      "step": 4645
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.62059127488021e-05,
+      "loss": 1.04,
+      "step": 4650
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.61919426081012e-05,
+      "loss": 1.0372,
+      "step": 4655
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.617794827453976e-05,
+      "loss": 1.0638,
+      "step": 4660
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.616392975897198e-05,
+      "loss": 1.0668,
+      "step": 4665
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.61498870722708e-05,
+      "loss": 1.0401,
+      "step": 4670
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.613582022532789e-05,
+      "loss": 1.0202,
+      "step": 4675
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.612172922905366e-05,
+      "loss": 1.0506,
+      "step": 4680
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.610761409437729e-05,
+      "loss": 1.0338,
+      "step": 4685
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.609347483224663e-05,
+      "loss": 1.0314,
+      "step": 4690
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.607931145362828e-05,
+      "loss": 1.003,
+      "step": 4695
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.606512396950754e-05,
+      "loss": 1.0124,
+      "step": 4700
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.605091239088839e-05,
+      "loss": 1.0458,
+      "step": 4705
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.603667672879352e-05,
+      "loss": 1.0338,
+      "step": 4710
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.602241699426427e-05,
+      "loss": 1.0204,
+      "step": 4715
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.600813319836068e-05,
+      "loss": 1.036,
+      "step": 4720
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 6.599382535216143e-05,
+      "loss": 1.0396,
+      "step": 4725
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 6.597949346676387e-05,
+      "loss": 1.0272,
+      "step": 4730
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 6.5965137553284e-05,
+      "loss": 1.019,
+      "step": 4735
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 6.595075762285644e-05,
+      "loss": 1.0041,
+      "step": 4740
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 6.593635368663444e-05,
+      "loss": 1.0533,
+      "step": 4745
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 6.592192575578988e-05,
+      "loss": 1.0264,
+      "step": 4750
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 6.590747384151322e-05,
+      "loss": 1.006,
+      "step": 4755
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 6.589299795501357e-05,
+      "loss": 1.0217,
+      "step": 4760
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 6.58784981075186e-05,
+      "loss": 1.0659,
+      "step": 4765
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 6.586397431027458e-05,
+      "loss": 1.0309,
+      "step": 4770
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 6.584942657454634e-05,
+      "loss": 1.0149,
+      "step": 4775
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 6.58348549116173e-05,
+      "loss": 1.0061,
+      "step": 4780
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 6.582025933278939e-05,
+      "loss": 1.0276,
+      "step": 4785
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.580563984938315e-05,
+      "loss": 1.0054,
+      "step": 4790
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.579099647273765e-05,
+      "loss": 1.0162,
+      "step": 4795
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.577632921421043e-05,
+      "loss": 0.9991,
+      "step": 4800
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.576163808517765e-05,
+      "loss": 1.0318,
+      "step": 4805
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.57469230970339e-05,
+      "loss": 1.0532,
+      "step": 4810
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.573218426119231e-05,
+      "loss": 1.0483,
+      "step": 4815
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.571742158908452e-05,
+      "loss": 1.0153,
+      "step": 4820
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.570263509216063e-05,
+      "loss": 1.0475,
+      "step": 4825
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.568782478188924e-05,
+      "loss": 1.0258,
+      "step": 4830
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.567299066975739e-05,
+      "loss": 1.0167,
+      "step": 4835
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.565813276727063e-05,
+      "loss": 1.0374,
+      "step": 4840
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.564325108595295e-05,
+      "loss": 1.0573,
+      "step": 4845
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.562834563734672e-05,
+      "loss": 1.0153,
+      "step": 4850
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.561341643301283e-05,
+      "loss": 1.0108,
+      "step": 4855
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.559846348453053e-05,
+      "loss": 1.0117,
+      "step": 4860
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.558348680349755e-05,
+      "loss": 1.0764,
+      "step": 4865
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.556848640152995e-05,
+      "loss": 1.0358,
+      "step": 4870
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.555346229026225e-05,
+      "loss": 1.0472,
+      "step": 4875
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.553841448134735e-05,
+      "loss": 0.995,
+      "step": 4880
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.55233429864565e-05,
+      "loss": 1.0324,
+      "step": 4885
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.550824781727934e-05,
+      "loss": 1.0377,
+      "step": 4890
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.549312898552386e-05,
+      "loss": 1.0086,
+      "step": 4895
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.547798650291646e-05,
+      "loss": 1.0191,
+      "step": 4900
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.546282038120179e-05,
+      "loss": 1.0607,
+      "step": 4905
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.544763063214291e-05,
+      "loss": 1.0465,
+      "step": 4910
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.543241726752117e-05,
+      "loss": 1.032,
+      "step": 4915
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.541718029913627e-05,
+      "loss": 1.0118,
+      "step": 4920
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.540191973880616e-05,
+      "loss": 1.0433,
+      "step": 4925
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.538663559836712e-05,
+      "loss": 1.0091,
+      "step": 4930
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.537132788967375e-05,
+      "loss": 1.0245,
+      "step": 4935
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.535599662459887e-05,
+      "loss": 1.0362,
+      "step": 4940
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.534064181503364e-05,
+      "loss": 1.0251,
+      "step": 4945
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.532526347288739e-05,
+      "loss": 1.0006,
+      "step": 4950
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.53098616100878e-05,
+      "loss": 1.0176,
+      "step": 4955
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.529443623858072e-05,
+      "loss": 1.037,
+      "step": 4960
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.527898737033025e-05,
+      "loss": 1.0143,
+      "step": 4965
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.526351501731878e-05,
+      "loss": 1.0159,
+      "step": 4970
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.524801919154681e-05,
+      "loss": 1.0218,
+      "step": 4975
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.52324999050331e-05,
+      "loss": 1.0517,
+      "step": 4980
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.521695716981463e-05,
+      "loss": 1.0134,
+      "step": 4985
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.52013909979465e-05,
+      "loss": 1.0321,
+      "step": 4990
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.518580140150208e-05,
+      "loss": 1.057,
+      "step": 4995
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.517018839257282e-05,
+      "loss": 1.0317,
+      "step": 5000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.515455198326837e-05,
+      "loss": 1.0322,
+      "step": 5005
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.513889218571653e-05,
+      "loss": 1.0648,
+      "step": 5010
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.512320901206322e-05,
+      "loss": 1.0306,
+      "step": 5015
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.510750247447252e-05,
+      "loss": 1.0492,
+      "step": 5020
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.50917725851266e-05,
+      "loss": 1.0343,
+      "step": 5025
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.50760193562258e-05,
+      "loss": 0.9919,
+      "step": 5030
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.506024279998848e-05,
+      "loss": 1.0478,
+      "step": 5035
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.504444292865115e-05,
+      "loss": 1.0566,
+      "step": 5040
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.502861975446838e-05,
+      "loss": 0.9775,
+      "step": 5045
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.501277328971283e-05,
+      "loss": 1.0211,
+      "step": 5050
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.499690354667521e-05,
+      "loss": 1.0223,
+      "step": 5055
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.49810105376643e-05,
+      "loss": 1.0719,
+      "step": 5060
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.496509427500691e-05,
+      "loss": 1.0188,
+      "step": 5065
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.494915477104788e-05,
+      "loss": 1.0111,
+      "step": 5070
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.493319203815011e-05,
+      "loss": 1.011,
+      "step": 5075
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.491720608869449e-05,
+      "loss": 1.0198,
+      "step": 5080
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.490119693507992e-05,
+      "loss": 0.9988,
+      "step": 5085
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.488516458972328e-05,
+      "loss": 1.0136,
+      "step": 5090
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.486910906505949e-05,
+      "loss": 0.9904,
+      "step": 5095
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.485303037354141e-05,
+      "loss": 1.0316,
+      "step": 5100
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.483692852763985e-05,
+      "loss": 1.0397,
+      "step": 5105
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.482080353984362e-05,
+      "loss": 1.0275,
+      "step": 5110
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.480465542265948e-05,
+      "loss": 1.0106,
+      "step": 5115
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.478848418861206e-05,
+      "loss": 0.9727,
+      "step": 5120
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.477228985024402e-05,
+      "loss": 1.039,
+      "step": 5125
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.475607242011587e-05,
+      "loss": 1.023,
+      "step": 5130
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.473983191080606e-05,
+      "loss": 0.9933,
+      "step": 5135
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.472356833491092e-05,
+      "loss": 0.9649,
+      "step": 5140
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.47072817050447e-05,
+      "loss": 1.0183,
+      "step": 5145
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.469097203383949e-05,
+      "loss": 1.0408,
+      "step": 5150
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.46746393339453e-05,
+      "loss": 0.9864,
+      "step": 5155
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.465828361802997e-05,
+      "loss": 0.9727,
+      "step": 5160
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.464190489877919e-05,
+      "loss": 1.0652,
+      "step": 5165
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.46255031888965e-05,
+      "loss": 1.0304,
+      "step": 5170
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.460907850110327e-05,
+      "loss": 1.0107,
+      "step": 5175
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.45926308481387e-05,
+      "loss": 0.9809,
+      "step": 5180
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.457616024275979e-05,
+      "loss": 1.037,
+      "step": 5185
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.455966669774134e-05,
+      "loss": 0.9996,
+      "step": 5190
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.454315022587598e-05,
+      "loss": 1.01,
+      "step": 5195
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.452661083997405e-05,
+      "loss": 1.0495,
+      "step": 5200
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.451004855286373e-05,
+      "loss": 1.0453,
+      "step": 5205
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.449346337739094e-05,
+      "loss": 1.017,
+      "step": 5210
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.447685532641932e-05,
+      "loss": 1.0165,
+      "step": 5215
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.446022441283034e-05,
+      "loss": 1.0016,
+      "step": 5220
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.444357064952308e-05,
+      "loss": 0.9909,
+      "step": 5225
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.442689404941444e-05,
+      "loss": 0.9959,
+      "step": 5230
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.441019462543899e-05,
+      "loss": 1.0517,
+      "step": 5235
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.439347239054902e-05,
+      "loss": 1.0253,
+      "step": 5240
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.437672735771451e-05,
+      "loss": 1.0436,
+      "step": 5245
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.43599595399231e-05,
+      "loss": 1.0097,
+      "step": 5250
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.434316895018012e-05,
+      "loss": 1.0301,
+      "step": 5255
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.432635560150857e-05,
+      "loss": 1.0116,
+      "step": 5260
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.43095195069491e-05,
+      "loss": 1.0094,
+      "step": 5265
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.429266067955998e-05,
+      "loss": 0.9927,
+      "step": 5270
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.427577913241713e-05,
+      "loss": 1.0411,
+      "step": 5275
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.42588748786141e-05,
+      "loss": 1.0225,
+      "step": 5280
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.424194793126203e-05,
+      "loss": 1.0346,
+      "step": 5285
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.422499830348967e-05,
+      "loss": 1.0307,
+      "step": 5290
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.420802600844337e-05,
+      "loss": 1.0496,
+      "step": 5295
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.419103105928706e-05,
+      "loss": 1.0001,
+      "step": 5300
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.417401346920222e-05,
+      "loss": 0.9814,
+      "step": 5305
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.415697325138792e-05,
+      "loss": 1.0326,
+      "step": 5310
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.413991041906074e-05,
+      "loss": 1.0164,
+      "step": 5315
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 6.412282498545486e-05,
+      "loss": 1.0209,
+      "step": 5320
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 6.410571696382193e-05,
+      "loss": 1.0289,
+      "step": 5325
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 6.408858636743117e-05,
+      "loss": 1.051,
+      "step": 5330
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 6.407143320956927e-05,
+      "loss": 1.037,
+      "step": 5335
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 6.405425750354042e-05,
+      "loss": 1.0062,
+      "step": 5340
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 6.403705926266633e-05,
+      "loss": 1.0349,
+      "step": 5345
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 6.401983850028617e-05,
+      "loss": 0.9925,
+      "step": 5350
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 6.400259522975658e-05,
+      "loss": 1.0669,
+      "step": 5355
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 6.398532946445164e-05,
+      "loss": 0.9948,
+      "step": 5360
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 6.396804121776292e-05,
+      "loss": 0.987,
+      "step": 5365
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 6.395073050309937e-05,
+      "loss": 1.0264,
+      "step": 5370
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 6.393339733388742e-05,
+      "loss": 1.0384,
+      "step": 5375
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 6.391604172357089e-05,
+      "loss": 1.0349,
+      "step": 5380
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 6.389866368561099e-05,
+      "loss": 1.0177,
+      "step": 5385
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 6.388126323348637e-05,
+      "loss": 1.0041,
+      "step": 5390
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 6.386384038069302e-05,
+      "loss": 1.0387,
+      "step": 5395
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 6.384639514074431e-05,
+      "loss": 0.9884,
+      "step": 5400
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 6.382892752717102e-05,
+      "loss": 0.9988,
+      "step": 5405
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 6.381143755352122e-05,
+      "loss": 1.0188,
+      "step": 5410
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 6.379392523336039e-05,
+      "loss": 1.0192,
+      "step": 5415
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 6.377639058027126e-05,
+      "loss": 1.0217,
+      "step": 5420
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 6.375883360785394e-05,
+      "loss": 1.041,
+      "step": 5425
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 6.374125432972584e-05,
+      "loss": 1.0139,
+      "step": 5430
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 6.372365275952169e-05,
+      "loss": 1.0473,
+      "step": 5435
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 6.370602891089344e-05,
+      "loss": 1.0598,
+      "step": 5440
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 6.368838279751042e-05,
+      "loss": 1.0299,
+      "step": 5445
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.367071443305915e-05,
+      "loss": 1.0093,
+      "step": 5450
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.365302383124342e-05,
+      "loss": 1.0485,
+      "step": 5455
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.36353110057843e-05,
+      "loss": 0.9474,
+      "step": 5460
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.361757597042009e-05,
+      "loss": 1.0163,
+      "step": 5465
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.359981873890627e-05,
+      "loss": 1.0179,
+      "step": 5470
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.358203932501562e-05,
+      "loss": 1.0407,
+      "step": 5475
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.356423774253803e-05,
+      "loss": 1.0328,
+      "step": 5480
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.354641400528064e-05,
+      "loss": 0.9768,
+      "step": 5485
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.352856812706778e-05,
+      "loss": 1.0601,
+      "step": 5490
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.351070012174093e-05,
+      "loss": 1.047,
+      "step": 5495
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.349281000315873e-05,
+      "loss": 1.0168,
+      "step": 5500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.347489778519698e-05,
+      "loss": 0.9969,
+      "step": 5505
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.345696348174863e-05,
+      "loss": 1.0376,
+      "step": 5510
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.343900710672374e-05,
+      "loss": 1.0101,
+      "step": 5515
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.342102867404952e-05,
+      "loss": 1.0402,
+      "step": 5520
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.340302819767023e-05,
+      "loss": 0.9698,
+      "step": 5525
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.338500569154731e-05,
+      "loss": 0.9928,
+      "step": 5530
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.33669611696592e-05,
+      "loss": 0.9698,
+      "step": 5535
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.334889464600146e-05,
+      "loss": 0.9902,
+      "step": 5540
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.333080613458674e-05,
+      "loss": 0.9941,
+      "step": 5545
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.33126956494447e-05,
+      "loss": 1.0108,
+      "step": 5550
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.329456320462205e-05,
+      "loss": 0.9757,
+      "step": 5555
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.327640881418256e-05,
+      "loss": 1.0253,
+      "step": 5560
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.325823249220699e-05,
+      "loss": 1.025,
+      "step": 5565
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.324003425279312e-05,
+      "loss": 0.9976,
+      "step": 5570
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.322181411005573e-05,
+      "loss": 0.9977,
+      "step": 5575
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.320357207812661e-05,
+      "loss": 1.0053,
+      "step": 5580
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.318530817115449e-05,
+      "loss": 1.0265,
+      "step": 5585
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.316702240330508e-05,
+      "loss": 1.0058,
+      "step": 5590
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.314871478876107e-05,
+      "loss": 1.0024,
+      "step": 5595
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.313038534172206e-05,
+      "loss": 1.0274,
+      "step": 5600
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.31120340764046e-05,
+      "loss": 1.0286,
+      "step": 5605
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.309366100704215e-05,
+      "loss": 0.9798,
+      "step": 5610
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.307526614788512e-05,
+      "loss": 1.0169,
+      "step": 5615
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.305684951320078e-05,
+      "loss": 1.0126,
+      "step": 5620
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.30384111172733e-05,
+      "loss": 0.9943,
+      "step": 5625
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.301995097440372e-05,
+      "loss": 0.9914,
+      "step": 5630
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.300146909890996e-05,
+      "loss": 0.9939,
+      "step": 5635
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.298296550512682e-05,
+      "loss": 0.9787,
+      "step": 5640
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.296444020740589e-05,
+      "loss": 1.0041,
+      "step": 5645
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.294589322011563e-05,
+      "loss": 1.031,
+      "step": 5650
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.292732455764132e-05,
+      "loss": 1.0136,
+      "step": 5655
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.290873423438504e-05,
+      "loss": 1.0044,
+      "step": 5660
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.289012226476568e-05,
+      "loss": 0.9805,
+      "step": 5665
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.287148866321893e-05,
+      "loss": 1.0401,
+      "step": 5670
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.285283344419722e-05,
+      "loss": 0.9893,
+      "step": 5675
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.283415662216978e-05,
+      "loss": 0.9691,
+      "step": 5680
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.28154582116226e-05,
+      "loss": 0.9968,
+      "step": 5685
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.279673822705838e-05,
+      "loss": 1.0057,
+      "step": 5690
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.277799668299659e-05,
+      "loss": 0.9683,
+      "step": 5695
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.27592335939734e-05,
+      "loss": 1.0042,
+      "step": 5700
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.27404489745417e-05,
+      "loss": 0.9661,
+      "step": 5705
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.272164283927106e-05,
+      "loss": 0.9715,
+      "step": 5710
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.270281520274776e-05,
+      "loss": 0.9801,
+      "step": 5715
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.268396607957477e-05,
+      "loss": 0.9954,
+      "step": 5720
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.26650954843717e-05,
+      "loss": 1.007,
+      "step": 5725
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.264620343177479e-05,
+      "loss": 0.998,
+      "step": 5730
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.262728993643697e-05,
+      "loss": 0.994,
+      "step": 5735
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.26083550130278e-05,
+      "loss": 0.9773,
+      "step": 5740
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.258939867623343e-05,
+      "loss": 0.9951,
+      "step": 5745
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.257042094075661e-05,
+      "loss": 1.0106,
+      "step": 5750
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.255142182131675e-05,
+      "loss": 1.0458,
+      "step": 5755
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.253240133264977e-05,
+      "loss": 0.9804,
+      "step": 5760
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.251335948950823e-05,
+      "loss": 0.9882,
+      "step": 5765
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.24942963066612e-05,
+      "loss": 0.9721,
+      "step": 5770
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.247521179889433e-05,
+      "loss": 1.0139,
+      "step": 5775
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.245610598100979e-05,
+      "loss": 1.0184,
+      "step": 5780
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.24369788678263e-05,
+      "loss": 0.9819,
+      "step": 5785
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.241783047417909e-05,
+      "loss": 0.9955,
+      "step": 5790
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.23986608149199e-05,
+      "loss": 0.9992,
+      "step": 5795
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.237946990491693e-05,
+      "loss": 1.0075,
+      "step": 5800
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.236025775905494e-05,
+      "loss": 0.975,
+      "step": 5805
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.234102439223506e-05,
+      "loss": 1.0778,
+      "step": 5810
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.232176981937496e-05,
+      "loss": 1.0063,
+      "step": 5815
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.230249405540871e-05,
+      "loss": 1.026,
+      "step": 5820
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.228319711528683e-05,
+      "loss": 1.0228,
+      "step": 5825
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.22638790139763e-05,
+      "loss": 0.9607,
+      "step": 5830
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.224453976646045e-05,
+      "loss": 1.0012,
+      "step": 5835
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.222517938773905e-05,
+      "loss": 1.0107,
+      "step": 5840
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.220579789282824e-05,
+      "loss": 0.9907,
+      "step": 5845
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.218639529676059e-05,
+      "loss": 0.9985,
+      "step": 5850
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.216697161458496e-05,
+      "loss": 1.0173,
+      "step": 5855
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.21475268613666e-05,
+      "loss": 0.9853,
+      "step": 5860
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.212806105218711e-05,
+      "loss": 0.9788,
+      "step": 5865
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.210857420214442e-05,
+      "loss": 1.0134,
+      "step": 5870
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.208906632635278e-05,
+      "loss": 1.0132,
+      "step": 5875
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.206953743994273e-05,
+      "loss": 0.9803,
+      "step": 5880
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.204998755806114e-05,
+      "loss": 1.0221,
+      "step": 5885
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.20304166958711e-05,
+      "loss": 0.982,
+      "step": 5890
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.201082486855205e-05,
+      "loss": 1.0084,
+      "step": 5895
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.199121209129965e-05,
+      "loss": 1.0187,
+      "step": 5900
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.197157837932582e-05,
+      "loss": 0.9765,
+      "step": 5905
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.195192374785869e-05,
+      "loss": 0.9952,
+      "step": 5910
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.193224821214267e-05,
+      "loss": 0.9985,
+      "step": 5915
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.191255178743832e-05,
+      "loss": 0.9814,
+      "step": 5920
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.189283448902245e-05,
+      "loss": 0.9993,
+      "step": 5925
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.187309633218803e-05,
+      "loss": 0.9883,
+      "step": 5930
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.185333733224424e-05,
+      "loss": 0.9774,
+      "step": 5935
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.18335575045164e-05,
+      "loss": 1.0324,
+      "step": 5940
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.181375686434599e-05,
+      "loss": 0.9906,
+      "step": 5945
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.179393542709064e-05,
+      "loss": 0.9984,
+      "step": 5950
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.177409320812409e-05,
+      "loss": 1.007,
+      "step": 5955
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.175423022283622e-05,
+      "loss": 1.0303,
+      "step": 5960
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.173434648663302e-05,
+      "loss": 1.0185,
+      "step": 5965
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.171444201493655e-05,
+      "loss": 1.0027,
+      "step": 5970
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.169451682318498e-05,
+      "loss": 1.0235,
+      "step": 5975
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 6.167457092683251e-05,
+      "loss": 1.0099,
+      "step": 5980
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 6.165460434134947e-05,
+      "loss": 0.9912,
+      "step": 5985
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 6.163461708222215e-05,
+      "loss": 0.9701,
+      "step": 5990
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 6.161460916495296e-05,
+      "loss": 0.9747,
+      "step": 5995
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 6.159458060506025e-05,
+      "loss": 1.007,
+      "step": 6000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 6.157453141807845e-05,
+      "loss": 0.9642,
+      "step": 6005
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 6.155446161955794e-05,
+      "loss": 0.9679,
+      "step": 6010
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 6.153437122506513e-05,
+      "loss": 1.0002,
+      "step": 6015
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 6.151426025018237e-05,
+      "loss": 0.9772,
+      "step": 6020
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 6.149412871050797e-05,
+      "loss": 0.9581,
+      "step": 6025
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 6.147397662165623e-05,
+      "loss": 1.0094,
+      "step": 6030
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 6.145380399925733e-05,
+      "loss": 0.9781,
+      "step": 6035
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 6.143361085895742e-05,
+      "loss": 0.9553,
+      "step": 6040
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.141339721641857e-05,
+      "loss": 0.9742,
+      "step": 6045
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.139316308731871e-05,
+      "loss": 1.0055,
+      "step": 6050
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.137290848735172e-05,
+      "loss": 1.0145,
+      "step": 6055
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.135263343222728e-05,
+      "loss": 0.9485,
+      "step": 6060
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.133233793767099e-05,
+      "loss": 0.9698,
+      "step": 6065
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.131202201942427e-05,
+      "loss": 1.0436,
+      "step": 6070
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.129168569324445e-05,
+      "loss": 1.0152,
+      "step": 6075
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.12713289749046e-05,
+      "loss": 1.0018,
+      "step": 6080
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.125095188019366e-05,
+      "loss": 1.005,
+      "step": 6085
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.123055442491635e-05,
+      "loss": 0.969,
+      "step": 6090
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.12101366248932e-05,
+      "loss": 0.9976,
+      "step": 6095
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.11896984959605e-05,
+      "loss": 0.9532,
+      "step": 6100
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.116924005397031e-05,
+      "loss": 0.9823,
+      "step": 6105
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.114876131479048e-05,
+      "loss": 0.9501,
+      "step": 6110
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.112826229430456e-05,
+      "loss": 0.9739,
+      "step": 6115
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.110774300841186e-05,
+      "loss": 0.9668,
+      "step": 6120
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.108720347302737e-05,
+      "loss": 0.9818,
+      "step": 6125
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.106664370408185e-05,
+      "loss": 0.9881,
+      "step": 6130
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.104606371752167e-05,
+      "loss": 0.9554,
+      "step": 6135
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.102546352930894e-05,
+      "loss": 1.0035,
+      "step": 6140
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.1004843155421445e-05,
+      "loss": 0.9613,
+      "step": 6145
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.098420261185257e-05,
+      "loss": 1.0042,
+      "step": 6150
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.096354191461141e-05,
+      "loss": 1.039,
+      "step": 6155
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.094286107972263e-05,
+      "loss": 0.9825,
+      "step": 6160
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.092216012322656e-05,
+      "loss": 0.9948,
+      "step": 6165
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.090143906117913e-05,
+      "loss": 0.9831,
+      "step": 6170
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.0880697909651836e-05,
+      "loss": 1.0165,
+      "step": 6175
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.0859936684731794e-05,
+      "loss": 0.9822,
+      "step": 6180
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.0839155402521646e-05,
+      "loss": 1.0405,
+      "step": 6185
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.081835407913963e-05,
+      "loss": 0.9818,
+      "step": 6190
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.079753273071952e-05,
+      "loss": 1.0055,
+      "step": 6195
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.077669137341059e-05,
+      "loss": 0.9524,
+      "step": 6200
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.075583002337767e-05,
+      "loss": 1.0111,
+      "step": 6205
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.073494869680109e-05,
+      "loss": 1.0198,
+      "step": 6210
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.071404740987666e-05,
+      "loss": 0.969,
+      "step": 6215
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.069312617881568e-05,
+      "loss": 1.0011,
+      "step": 6220
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.067218501984492e-05,
+      "loss": 0.9849,
+      "step": 6225
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.065122394920661e-05,
+      "loss": 0.9879,
+      "step": 6230
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.063024298315841e-05,
+      "loss": 1.0061,
+      "step": 6235
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.060924213797341e-05,
+      "loss": 0.997,
+      "step": 6240
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.0588221429940156e-05,
+      "loss": 1.003,
+      "step": 6245
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.056718087536255e-05,
+      "loss": 0.9546,
+      "step": 6250
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.054612049055992e-05,
+      "loss": 0.9871,
+      "step": 6255
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.052504029186696e-05,
+      "loss": 1.0045,
+      "step": 6260
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.050394029563372e-05,
+      "loss": 1.0082,
+      "step": 6265
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.0482820518225644e-05,
+      "loss": 1.0085,
+      "step": 6270
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.046168097602349e-05,
+      "loss": 1.0118,
+      "step": 6275
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.044052168542334e-05,
+      "loss": 0.9625,
+      "step": 6280
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.0419342662836615e-05,
+      "loss": 0.9875,
+      "step": 6285
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.039814392469003e-05,
+      "loss": 0.9807,
+      "step": 6290
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.037692548742558e-05,
+      "loss": 0.9654,
+      "step": 6295
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.0355687367500565e-05,
+      "loss": 0.9182,
+      "step": 6300
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.033442958138753e-05,
+      "loss": 0.9512,
+      "step": 6305
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.031315214557428e-05,
+      "loss": 0.9459,
+      "step": 6310
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.029185507656385e-05,
+      "loss": 0.9476,
+      "step": 6315
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.027053839087452e-05,
+      "loss": 0.998,
+      "step": 6320
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.024920210503978e-05,
+      "loss": 0.9981,
+      "step": 6325
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.022784623560833e-05,
+      "loss": 0.9745,
+      "step": 6330
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.020647079914404e-05,
+      "loss": 0.9799,
+      "step": 6335
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.018507581222596e-05,
+      "loss": 0.9694,
+      "step": 6340
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.0163661291448304e-05,
+      "loss": 1.0028,
+      "step": 6345
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.014222725342046e-05,
+      "loss": 1.0295,
+      "step": 6350
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.0120773714766925e-05,
+      "loss": 0.9792,
+      "step": 6355
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.009930069212734e-05,
+      "loss": 0.9862,
+      "step": 6360
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.007780820215643e-05,
+      "loss": 1.0313,
+      "step": 6365
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.005629626152407e-05,
+      "loss": 0.9876,
+      "step": 6370
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.003476488691516e-05,
+      "loss": 0.9728,
+      "step": 6375
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.0013214095029726e-05,
+      "loss": 0.9436,
+      "step": 6380
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.999164390258281e-05,
+      "loss": 0.9614,
+      "step": 6385
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.9970054326304546e-05,
+      "loss": 1.0066,
+      "step": 6390
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.994844538294006e-05,
+      "loss": 0.9476,
+      "step": 6395
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.992681708924953e-05,
+      "loss": 0.9597,
+      "step": 6400
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.990516946200812e-05,
+      "loss": 0.9987,
+      "step": 6405
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.988350251800601e-05,
+      "loss": 0.9798,
+      "step": 6410
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.9861816274048356e-05,
+      "loss": 0.9642,
+      "step": 6415
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.984011074695527e-05,
+      "loss": 1.059,
+      "step": 6420
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.981838595356183e-05,
+      "loss": 0.9593,
+      "step": 6425
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.9796641910718076e-05,
+      "loss": 0.9706,
+      "step": 6430
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.977487863528893e-05,
+      "loss": 0.9171,
+      "step": 6435
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.975309614415427e-05,
+      "loss": 1.002,
+      "step": 6440
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.973129445420889e-05,
+      "loss": 0.9998,
+      "step": 6445
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.970947358236242e-05,
+      "loss": 0.9776,
+      "step": 6450
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.9687633545539414e-05,
+      "loss": 1.0045,
+      "step": 6455
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.96657743606793e-05,
+      "loss": 0.9935,
+      "step": 6460
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.964389604473631e-05,
+      "loss": 1.0191,
+      "step": 6465
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.962199861467954e-05,
+      "loss": 0.9951,
+      "step": 6470
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.9600082087492916e-05,
+      "loss": 0.9664,
+      "step": 6475
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.957814648017517e-05,
+      "loss": 0.9492,
+      "step": 6480
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.9556191809739834e-05,
+      "loss": 0.9713,
+      "step": 6485
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.953421809321522e-05,
+      "loss": 0.9854,
+      "step": 6490
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.951222534764443e-05,
+      "loss": 0.9842,
+      "step": 6495
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.94902135900853e-05,
+      "loss": 0.9735,
+      "step": 6500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.946818283761043e-05,
+      "loss": 0.9775,
+      "step": 6505
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.944613310730713e-05,
+      "loss": 0.9589,
+      "step": 6510
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.942406441627748e-05,
+      "loss": 1.008,
+      "step": 6515
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.94019767816382e-05,
+      "loss": 0.9766,
+      "step": 6520
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.9379870220520754e-05,
+      "loss": 0.9397,
+      "step": 6525
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.935774475007126e-05,
+      "loss": 0.9706,
+      "step": 6530
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.933560038745052e-05,
+      "loss": 0.9677,
+      "step": 6535
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.9313437149833964e-05,
+      "loss": 0.9712,
+      "step": 6540
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.929125505441168e-05,
+      "loss": 0.9412,
+      "step": 6545
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.9269054118388384e-05,
+      "loss": 0.9991,
+      "step": 6550
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.9246834358983395e-05,
+      "loss": 1.0212,
+      "step": 6555
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.922459579343064e-05,
+      "loss": 0.9533,
+      "step": 6560
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.920233843897862e-05,
+      "loss": 0.9427,
+      "step": 6565
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.918006231289042e-05,
+      "loss": 1.0162,
+      "step": 6570
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.915776743244369e-05,
+      "loss": 1.0145,
+      "step": 6575
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.913545381493061e-05,
+      "loss": 0.9546,
+      "step": 6580
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.9113121477657904e-05,
+      "loss": 0.9407,
+      "step": 6585
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.909077043794681e-05,
+      "loss": 0.9798,
+      "step": 6590
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.9068400713133076e-05,
+      "loss": 0.9725,
+      "step": 6595
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.9046012320566933e-05,
+      "loss": 0.9692,
+      "step": 6600
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.902360527761312e-05,
+      "loss": 1.038,
+      "step": 6605
+    },
+    {
+      "epoch": 1.0,
+      "eval_loss": 1.0505504608154297,
+      "eval_runtime": 235.5792,
+      "eval_samples_per_second": 47.233,
+      "eval_steps_per_second": 5.905,
+      "step": 6606
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.900117960165079e-05,
+      "loss": 0.9534,
+      "step": 6610
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.89787353100736e-05,
+      "loss": 0.8824,
+      "step": 6615
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.8956272420289603e-05,
+      "loss": 0.9111,
+      "step": 6620
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.893379094972132e-05,
+      "loss": 0.9253,
+      "step": 6625
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.891129091580564e-05,
+      "loss": 0.8864,
+      "step": 6630
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.888877233599389e-05,
+      "loss": 0.9017,
+      "step": 6635
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.8866235227751735e-05,
+      "loss": 0.9234,
+      "step": 6640
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.884367960855926e-05,
+      "loss": 0.9303,
+      "step": 6645
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.882110549591089e-05,
+      "loss": 0.9147,
+      "step": 6650
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.879851290731536e-05,
+      "loss": 0.9304,
+      "step": 6655
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.877590186029579e-05,
+      "loss": 0.8621,
+      "step": 6660
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.875327237238959e-05,
+      "loss": 0.9291,
+      "step": 6665
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.873062446114846e-05,
+      "loss": 0.944,
+      "step": 6670
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.870795814413842e-05,
+      "loss": 0.9013,
+      "step": 6675
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.8685273438939715e-05,
+      "loss": 0.8909,
+      "step": 6680
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.8662570363146925e-05,
+      "loss": 0.9104,
+      "step": 6685
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.8639848934368826e-05,
+      "loss": 0.944,
+      "step": 6690
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.861710917022843e-05,
+      "loss": 0.8816,
+      "step": 6695
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.8594351088362994e-05,
+      "loss": 0.9141,
+      "step": 6700
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.857157470642397e-05,
+      "loss": 0.9536,
+      "step": 6705
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.854878004207701e-05,
+      "loss": 0.9306,
+      "step": 6710
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.8525967113001925e-05,
+      "loss": 0.9299,
+      "step": 6715
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.8503135936892716e-05,
+      "loss": 0.9061,
+      "step": 6720
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.848028653145753e-05,
+      "loss": 0.8962,
+      "step": 6725
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.845741891441866e-05,
+      "loss": 0.9182,
+      "step": 6730
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.8434533103512496e-05,
+      "loss": 0.9587,
+      "step": 6735
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.841162911648957e-05,
+      "loss": 0.9212,
+      "step": 6740
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.83887069711145e-05,
+      "loss": 0.9174,
+      "step": 6745
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.8365766685166e-05,
+      "loss": 0.9278,
+      "step": 6750
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.8342808276436814e-05,
+      "loss": 0.8999,
+      "step": 6755
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.83198317627338e-05,
+      "loss": 0.9038,
+      "step": 6760
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.829683716187781e-05,
+      "loss": 0.91,
+      "step": 6765
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.827382449170375e-05,
+      "loss": 0.9069,
+      "step": 6770
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 5.8250793770060524e-05,
+      "loss": 0.9242,
+      "step": 6775
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 5.822774501481108e-05,
+      "loss": 0.9141,
+      "step": 6780
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 5.8204678243832286e-05,
+      "loss": 0.936,
+      "step": 6785
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 5.818159347501503e-05,
+      "loss": 0.9331,
+      "step": 6790
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 5.815849072626416e-05,
+      "loss": 0.8545,
+      "step": 6795
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 5.813537001549842e-05,
+      "loss": 0.9107,
+      "step": 6800
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 5.811223136065057e-05,
+      "loss": 0.9043,
+      "step": 6805
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 5.808907477966722e-05,
+      "loss": 0.8898,
+      "step": 6810
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 5.8065900290508885e-05,
+      "loss": 0.9563,
+      "step": 6815
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 5.804270791115001e-05,
+      "loss": 0.9006,
+      "step": 6820
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 5.8019497659578885e-05,
+      "loss": 0.8957,
+      "step": 6825
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 5.7996269553797684e-05,
+      "loss": 0.8813,
+      "step": 6830
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 5.79730236118224e-05,
+      "loss": 0.9058,
+      "step": 6835
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 5.794975985168288e-05,
+      "loss": 0.8922,
+      "step": 6840
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 5.792647829142279e-05,
+      "loss": 0.9023,
+      "step": 6845
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 5.790317894909961e-05,
+      "loss": 0.8944,
+      "step": 6850
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 5.7879861842784584e-05,
+      "loss": 0.893,
+      "step": 6855
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 5.785652699056275e-05,
+      "loss": 0.9061,
+      "step": 6860
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 5.783317441053293e-05,
+      "loss": 0.923,
+      "step": 6865
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 5.780980412080766e-05,
+      "loss": 0.931,
+      "step": 6870
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 5.7786416139513235e-05,
+      "loss": 0.8903,
+      "step": 6875
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 5.776301048478966e-05,
+      "loss": 0.9155,
+      "step": 6880
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 5.773958717479066e-05,
+      "loss": 0.9434,
+      "step": 6885
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 5.771614622768365e-05,
+      "loss": 0.9035,
+      "step": 6890
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 5.769268766164972e-05,
+      "loss": 0.864,
+      "step": 6895
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 5.766921149488361e-05,
+      "loss": 0.9216,
+      "step": 6900
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 5.764571774559375e-05,
+      "loss": 0.9276,
+      "step": 6905
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 5.7622206432002173e-05,
+      "loss": 0.8776,
+      "step": 6910
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 5.7598677572344545e-05,
+      "loss": 0.9265,
+      "step": 6915
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 5.757513118487016e-05,
+      "loss": 0.9297,
+      "step": 6920
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 5.755156728784188e-05,
+      "loss": 0.9513,
+      "step": 6925
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 5.752798589953617e-05,
+      "loss": 0.8794,
+      "step": 6930
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 5.750438703824302e-05,
+      "loss": 0.9036,
+      "step": 6935
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 5.748077072226603e-05,
+      "loss": 0.9382,
+      "step": 6940
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 5.74571369699223e-05,
+      "loss": 0.9321,
+      "step": 6945
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 5.743348579954246e-05,
+      "loss": 0.9127,
+      "step": 6950
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 5.740981722947066e-05,
+      "loss": 0.9217,
+      "step": 6955
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 5.738613127806453e-05,
+      "loss": 0.9219,
+      "step": 6960
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 5.73624279636952e-05,
+      "loss": 0.8869,
+      "step": 6965
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 5.733870730474727e-05,
+      "loss": 0.8838,
+      "step": 6970
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 5.731496931961874e-05,
+      "loss": 0.8702,
+      "step": 6975
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 5.729121402672112e-05,
+      "loss": 0.8991,
+      "step": 6980
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 5.726744144447929e-05,
+      "loss": 0.9338,
+      "step": 6985
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 5.724365159133159e-05,
+      "loss": 0.9069,
+      "step": 6990
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 5.7219844485729706e-05,
+      "loss": 0.9352,
+      "step": 6995
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 5.719602014613873e-05,
+      "loss": 0.9045,
+      "step": 7000
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 5.717217859103713e-05,
+      "loss": 0.8921,
+      "step": 7005
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 5.714831983891669e-05,
+      "loss": 0.9381,
+      "step": 7010
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 5.712444390828257e-05,
+      "loss": 0.9538,
+      "step": 7015
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 5.710055081765325e-05,
+      "loss": 0.9202,
+      "step": 7020
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 5.70766405855605e-05,
+      "loss": 0.9005,
+      "step": 7025
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 5.705271323054939e-05,
+      "loss": 0.9318,
+      "step": 7030
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 5.7028768771178285e-05,
+      "loss": 0.8939,
+      "step": 7035
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 5.7004807226018795e-05,
+      "loss": 0.8884,
+      "step": 7040
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 5.6980828613655805e-05,
+      "loss": 0.9003,
+      "step": 7045
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 5.695683295268744e-05,
+      "loss": 0.9471,
+      "step": 7050
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 5.6932820261724994e-05,
+      "loss": 0.9155,
+      "step": 7055
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 5.6908790559393046e-05,
+      "loss": 0.9303,
+      "step": 7060
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 5.688474386432931e-05,
+      "loss": 0.9219,
+      "step": 7065
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 5.6860680195184724e-05,
+      "loss": 0.9016,
+      "step": 7070
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 5.6836599570623363e-05,
+      "loss": 0.8859,
+      "step": 7075
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 5.681250200932245e-05,
+      "loss": 0.894,
+      "step": 7080
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 5.678838752997237e-05,
+      "loss": 0.9245,
+      "step": 7085
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 5.676425615127661e-05,
+      "loss": 0.9031,
+      "step": 7090
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 5.6740107891951757e-05,
+      "loss": 0.871,
+      "step": 7095
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 5.6715942770727515e-05,
+      "loss": 0.9115,
+      "step": 7100
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 5.669176080634664e-05,
+      "loss": 0.9441,
+      "step": 7105
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 5.666756201756499e-05,
+      "loss": 0.9037,
+      "step": 7110
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 5.664334642315144e-05,
+      "loss": 0.9654,
+      "step": 7115
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 5.66191140418879e-05,
+      "loss": 0.9235,
+      "step": 7120
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 5.659486489256931e-05,
+      "loss": 0.9104,
+      "step": 7125
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 5.657059899400361e-05,
+      "loss": 0.86,
+      "step": 7130
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 5.654631636501175e-05,
+      "loss": 0.8997,
+      "step": 7135
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 5.6522017024427615e-05,
+      "loss": 0.9169,
+      "step": 7140
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 5.649770099109809e-05,
+      "loss": 0.9034,
+      "step": 7145
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 5.6473368283883004e-05,
+      "loss": 0.9087,
+      "step": 7150
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 5.644901892165509e-05,
+      "loss": 0.9106,
+      "step": 7155
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 5.6424652923300035e-05,
+      "loss": 0.8709,
+      "step": 7160
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 5.6400270307716405e-05,
+      "loss": 0.8522,
+      "step": 7165
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 5.637587109381566e-05,
+      "loss": 0.9109,
+      "step": 7170
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 5.635145530052214e-05,
+      "loss": 0.9181,
+      "step": 7175
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 5.6327022946773026e-05,
+      "loss": 0.911,
+      "step": 7180
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 5.630257405151837e-05,
+      "loss": 0.9054,
+      "step": 7185
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 5.627810863372104e-05,
+      "loss": 0.8591,
+      "step": 7190
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 5.625362671235672e-05,
+      "loss": 0.9058,
+      "step": 7195
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 5.622912830641388e-05,
+      "loss": 0.9507,
+      "step": 7200
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 5.620461343489381e-05,
+      "loss": 0.9296,
+      "step": 7205
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 5.618008211681054e-05,
+      "loss": 0.913,
+      "step": 7210
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 5.615553437119087e-05,
+      "loss": 0.9396,
+      "step": 7215
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 5.613097021707434e-05,
+      "loss": 0.9179,
+      "step": 7220
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 5.6106389673513205e-05,
+      "loss": 0.9219,
+      "step": 7225
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 5.6081792759572446e-05,
+      "loss": 0.9807,
+      "step": 7230
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 5.605717949432975e-05,
+      "loss": 0.8857,
+      "step": 7235
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 5.6032549896875466e-05,
+      "loss": 0.9097,
+      "step": 7240
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 5.60079039863126e-05,
+      "loss": 0.8631,
+      "step": 7245
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 5.598324178175685e-05,
+      "loss": 0.903,
+      "step": 7250
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 5.595856330233653e-05,
+      "loss": 0.891,
+      "step": 7255
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 5.5933868567192556e-05,
+      "loss": 0.9367,
+      "step": 7260
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 5.590915759547849e-05,
+      "loss": 0.8877,
+      "step": 7265
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 5.588443040636047e-05,
+      "loss": 0.8926,
+      "step": 7270
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 5.585968701901721e-05,
+      "loss": 0.9178,
+      "step": 7275
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 5.583492745263999e-05,
+      "loss": 0.9155,
+      "step": 7280
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 5.581015172643264e-05,
+      "loss": 0.8947,
+      "step": 7285
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 5.5785359859611526e-05,
+      "loss": 0.931,
+      "step": 7290
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 5.5760551871405536e-05,
+      "loss": 0.8955,
+      "step": 7295
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 5.573572778105603e-05,
+      "loss": 0.9508,
+      "step": 7300
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 5.571088760781691e-05,
+      "loss": 0.879,
+      "step": 7305
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 5.56860313709545e-05,
+      "loss": 0.9025,
+      "step": 7310
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 5.5661159089747636e-05,
+      "loss": 0.9208,
+      "step": 7315
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 5.563627078348755e-05,
+      "loss": 0.9245,
+      "step": 7320
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 5.561136647147793e-05,
+      "loss": 0.8912,
+      "step": 7325
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 5.558644617303488e-05,
+      "loss": 0.9213,
+      "step": 7330
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 5.556150990748688e-05,
+      "loss": 0.8952,
+      "step": 7335
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 5.553655769417481e-05,
+      "loss": 0.9086,
+      "step": 7340
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 5.5511589552451926e-05,
+      "loss": 0.9381,
+      "step": 7345
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 5.5486605501683824e-05,
+      "loss": 0.9489,
+      "step": 7350
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 5.546160556124847e-05,
+      "loss": 0.8889,
+      "step": 7355
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 5.5436589750536095e-05,
+      "loss": 0.9193,
+      "step": 7360
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 5.541155808894929e-05,
+      "loss": 0.9055,
+      "step": 7365
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 5.538651059590292e-05,
+      "loss": 0.8767,
+      "step": 7370
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 5.536144729082415e-05,
+      "loss": 0.9158,
+      "step": 7375
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 5.533636819315238e-05,
+      "loss": 0.9182,
+      "step": 7380
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 5.531127332233926e-05,
+      "loss": 0.9154,
+      "step": 7385
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 5.528616269784871e-05,
+      "loss": 0.865,
+      "step": 7390
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 5.526103633915684e-05,
+      "loss": 0.9508,
+      "step": 7395
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 5.5235894265751947e-05,
+      "loss": 0.9149,
+      "step": 7400
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 5.5210736497134555e-05,
+      "loss": 0.9044,
+      "step": 7405
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 5.518556305281734e-05,
+      "loss": 0.9338,
+      "step": 7410
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 5.516037395232514e-05,
+      "loss": 0.8947,
+      "step": 7415
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 5.513516921519493e-05,
+      "loss": 0.9041,
+      "step": 7420
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 5.510994886097583e-05,
+      "loss": 0.8542,
+      "step": 7425
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 5.508471290922905e-05,
+      "loss": 0.907,
+      "step": 7430
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 5.505946137952793e-05,
+      "loss": 0.9079,
+      "step": 7435
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 5.503419429145785e-05,
+      "loss": 0.9465,
+      "step": 7440
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 5.5008911664616275e-05,
+      "loss": 0.9259,
+      "step": 7445
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 5.498361351861276e-05,
+      "loss": 0.9123,
+      "step": 7450
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 5.495829987306884e-05,
+      "loss": 0.9065,
+      "step": 7455
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 5.4932970747618075e-05,
+      "loss": 0.9244,
+      "step": 7460
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 5.4907626161906086e-05,
+      "loss": 0.8623,
+      "step": 7465
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 5.488226613559045e-05,
+      "loss": 0.8797,
+      "step": 7470
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 5.48568906883407e-05,
+      "loss": 0.8623,
+      "step": 7475
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 5.4831499839838353e-05,
+      "loss": 0.9219,
+      "step": 7480
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 5.480609360977688e-05,
+      "loss": 0.912,
+      "step": 7485
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 5.478067201786166e-05,
+      "loss": 0.9262,
+      "step": 7490
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 5.4755235083809996e-05,
+      "loss": 0.8968,
+      "step": 7495
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 5.47297828273511e-05,
+      "loss": 0.9141,
+      "step": 7500
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 5.470431526822607e-05,
+      "loss": 0.8791,
+      "step": 7505
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 5.467883242618784e-05,
+      "loss": 0.9156,
+      "step": 7510
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 5.4653334321001235e-05,
+      "loss": 0.9281,
+      "step": 7515
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 5.4627820972442904e-05,
+      "loss": 0.9188,
+      "step": 7520
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 5.460229240030131e-05,
+      "loss": 0.9078,
+      "step": 7525
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 5.457674862437674e-05,
+      "loss": 0.9114,
+      "step": 7530
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 5.4551189664481264e-05,
+      "loss": 0.9215,
+      "step": 7535
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 5.452561554043874e-05,
+      "loss": 0.8958,
+      "step": 7540
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 5.450002627208476e-05,
+      "loss": 0.9083,
+      "step": 7545
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 5.4474421879266676e-05,
+      "loss": 0.9247,
+      "step": 7550
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 5.444880238184361e-05,
+      "loss": 0.9541,
+      "step": 7555
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 5.4423167799686334e-05,
+      "loss": 0.8755,
+      "step": 7560
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 5.439751815267736e-05,
+      "loss": 0.9032,
+      "step": 7565
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 5.4371853460710875e-05,
+      "loss": 0.919,
+      "step": 7570
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 5.434617374369273e-05,
+      "loss": 0.9392,
+      "step": 7575
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 5.432047902154044e-05,
+      "loss": 0.8893,
+      "step": 7580
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 5.429476931418315e-05,
+      "loss": 0.895,
+      "step": 7585
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 5.426904464156164e-05,
+      "loss": 0.9074,
+      "step": 7590
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 5.424330502362827e-05,
+      "loss": 0.9394,
+      "step": 7595
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 5.421755048034703e-05,
+      "loss": 0.8899,
+      "step": 7600
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 5.4191781031693445e-05,
+      "loss": 0.8983,
+      "step": 7605
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 5.4165996697654627e-05,
+      "loss": 0.9945,
+      "step": 7610
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 5.4140197498229224e-05,
+      "loss": 0.8803,
+      "step": 7615
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 5.411438345342744e-05,
+      "loss": 0.8916,
+      "step": 7620
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 5.4088554583270957e-05,
+      "loss": 0.9153,
+      "step": 7625
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 5.406271090779295e-05,
+      "loss": 0.896,
+      "step": 7630
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 5.403685244703813e-05,
+      "loss": 0.8826,
+      "step": 7635
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 5.4010979221062634e-05,
+      "loss": 0.8899,
+      "step": 7640
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 5.398509124993404e-05,
+      "loss": 0.9053,
+      "step": 7645
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 5.395918855373139e-05,
+      "loss": 0.94,
+      "step": 7650
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 5.393327115254515e-05,
+      "loss": 0.9014,
+      "step": 7655
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 5.390733906647718e-05,
+      "loss": 0.8912,
+      "step": 7660
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 5.388139231564071e-05,
+      "loss": 0.9102,
+      "step": 7665
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 5.385543092016038e-05,
+      "loss": 0.8637,
+      "step": 7670
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 5.382945490017218e-05,
+      "loss": 0.86,
+      "step": 7675
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 5.380346427582342e-05,
+      "loss": 0.8906,
+      "step": 7680
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 5.3777459067272765e-05,
+      "loss": 0.9093,
+      "step": 7685
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 5.375143929469017e-05,
+      "loss": 0.9411,
+      "step": 7690
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 5.37254049782569e-05,
+      "loss": 0.8865,
+      "step": 7695
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 5.3699356138165496e-05,
+      "loss": 0.9028,
+      "step": 7700
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 5.367329279461977e-05,
+      "loss": 0.8627,
+      "step": 7705
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 5.364721496783476e-05,
+      "loss": 0.8653,
+      "step": 7710
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 5.3621122678036777e-05,
+      "loss": 0.8904,
+      "step": 7715
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 5.359501594546332e-05,
+      "loss": 0.9245,
+      "step": 7720
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 5.356889479036307e-05,
+      "loss": 0.8765,
+      "step": 7725
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 5.3542759232995954e-05,
+      "loss": 0.8737,
+      "step": 7730
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 5.351660929363303e-05,
+      "loss": 0.8931,
+      "step": 7735
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 5.3490444992556534e-05,
+      "loss": 0.9353,
+      "step": 7740
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 5.3464266350059773e-05,
+      "loss": 0.9068,
+      "step": 7745
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 5.343807338644728e-05,
+      "loss": 0.8703,
+      "step": 7750
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 5.341186612203462e-05,
+      "loss": 0.8871,
+      "step": 7755
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 5.3385644577148484e-05,
+      "loss": 0.8746,
+      "step": 7760
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 5.335940877212664e-05,
+      "loss": 0.8787,
+      "step": 7765
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 5.333315872731787e-05,
+      "loss": 0.8642,
+      "step": 7770
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 5.330689446308208e-05,
+      "loss": 0.8905,
+      "step": 7775
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 5.3280615999790136e-05,
+      "loss": 0.908,
+      "step": 7780
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 5.325432335782395e-05,
+      "loss": 0.8939,
+      "step": 7785
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 5.322801655757642e-05,
+      "loss": 0.9057,
+      "step": 7790
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 5.320169561945143e-05,
+      "loss": 0.8556,
+      "step": 7795
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 5.317536056386384e-05,
+      "loss": 0.951,
+      "step": 7800
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 5.3149011411239435e-05,
+      "loss": 0.9084,
+      "step": 7805
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 5.312264818201494e-05,
+      "loss": 0.8865,
+      "step": 7810
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 5.309627089663803e-05,
+      "loss": 0.877,
+      "step": 7815
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 5.306987957556724e-05,
+      "loss": 0.8925,
+      "step": 7820
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 5.3043474239272005e-05,
+      "loss": 0.858,
+      "step": 7825
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 5.301705490823264e-05,
+      "loss": 0.8915,
+      "step": 7830
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 5.299062160294031e-05,
+      "loss": 0.8778,
+      "step": 7835
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 5.296417434389702e-05,
+      "loss": 0.9101,
+      "step": 7840
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 5.293771315161559e-05,
+      "loss": 0.8882,
+      "step": 7845
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 5.291123804661965e-05,
+      "loss": 0.9182,
+      "step": 7850
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 5.288474904944363e-05,
+      "loss": 0.9319,
+      "step": 7855
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 5.285824618063272e-05,
+      "loss": 0.9254,
+      "step": 7860
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 5.283172946074287e-05,
+      "loss": 0.9075,
+      "step": 7865
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 5.280519891034079e-05,
+      "loss": 0.9215,
+      "step": 7870
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 5.2778654550003894e-05,
+      "loss": 0.9285,
+      "step": 7875
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 5.275209640032035e-05,
+      "loss": 0.9072,
+      "step": 7880
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 5.2725524481888964e-05,
+      "loss": 0.8713,
+      "step": 7885
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 5.2698938815319236e-05,
+      "loss": 0.8701,
+      "step": 7890
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 5.267233942123137e-05,
+      "loss": 0.8915,
+      "step": 7895
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 5.2645726320256186e-05,
+      "loss": 0.8991,
+      "step": 7900
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 5.261909953303512e-05,
+      "loss": 0.8695,
+      "step": 7905
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 5.259245908022026e-05,
+      "loss": 0.9014,
+      "step": 7910
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 5.256580498247425e-05,
+      "loss": 0.9184,
+      "step": 7915
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 5.253913726047037e-05,
+      "loss": 0.8743,
+      "step": 7920
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 5.251245593489241e-05,
+      "loss": 0.872,
+      "step": 7925
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 5.248576102643477e-05,
+      "loss": 0.922,
+      "step": 7930
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 5.245905255580232e-05,
+      "loss": 0.9029,
+      "step": 7935
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 5.2432330543710515e-05,
+      "loss": 0.8999,
+      "step": 7940
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 5.240559501088527e-05,
+      "loss": 0.884,
+      "step": 7945
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 5.237884597806299e-05,
+      "loss": 0.9094,
+      "step": 7950
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 5.235208346599057e-05,
+      "loss": 0.92,
+      "step": 7955
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 5.232530749542535e-05,
+      "loss": 0.8884,
+      "step": 7960
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 5.2298518087135096e-05,
+      "loss": 0.9118,
+      "step": 7965
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 5.227171526189801e-05,
+      "loss": 0.9027,
+      "step": 7970
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 5.224489904050271e-05,
+      "loss": 0.9177,
+      "step": 7975
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 5.2218069443748183e-05,
+      "loss": 0.935,
+      "step": 7980
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 5.21912264924438e-05,
+      "loss": 0.8696,
+      "step": 7985
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 5.21643702074093e-05,
+      "loss": 0.8825,
+      "step": 7990
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 5.213750060947473e-05,
+      "loss": 0.914,
+      "step": 7995
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 5.211061771948052e-05,
+      "loss": 0.8782,
+      "step": 8000
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 5.2083721558277344e-05,
+      "loss": 0.8953,
+      "step": 8005
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 5.205681214672621e-05,
+      "loss": 0.8765,
+      "step": 8010
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 5.202988950569839e-05,
+      "loss": 0.9045,
+      "step": 8015
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 5.200295365607545e-05,
+      "loss": 0.8877,
+      "step": 8020
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 5.197600461874914e-05,
+      "loss": 0.846,
+      "step": 8025
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 5.194904241462147e-05,
+      "loss": 0.9097,
+      "step": 8030
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 5.1922067064604685e-05,
+      "loss": 0.9371,
+      "step": 8035
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 5.1895078589621177e-05,
+      "loss": 0.9815,
+      "step": 8040
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 5.186807701060357e-05,
+      "loss": 0.8959,
+      "step": 8045
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 5.18410623484946e-05,
+      "loss": 0.9093,
+      "step": 8050
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 5.1814034624247204e-05,
+      "loss": 0.8645,
+      "step": 8055
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 5.17869938588244e-05,
+      "loss": 0.8725,
+      "step": 8060
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 5.175994007319936e-05,
+      "loss": 0.866,
+      "step": 8065
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 5.173287328835534e-05,
+      "loss": 0.8949,
+      "step": 8070
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 5.170579352528565e-05,
+      "loss": 0.8709,
+      "step": 8075
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 5.1678700804993717e-05,
+      "loss": 0.9111,
+      "step": 8080
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 5.1651595148492985e-05,
+      "loss": 0.868,
+      "step": 8085
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 5.1624476576806955e-05,
+      "loss": 0.9123,
+      "step": 8090
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 5.15973451109691e-05,
+      "loss": 0.9473,
+      "step": 8095
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 5.157020077202295e-05,
+      "loss": 0.9171,
+      "step": 8100
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 5.154304358102198e-05,
+      "loss": 0.905,
+      "step": 8105
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 5.1515873559029655e-05,
+      "loss": 0.8761,
+      "step": 8110
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 5.148869072711939e-05,
+      "loss": 0.8952,
+      "step": 8115
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 5.146149510637449e-05,
+      "loss": 0.9485,
+      "step": 8120
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 5.1434286717888285e-05,
+      "loss": 0.9087,
+      "step": 8125
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 5.1407065582763874e-05,
+      "loss": 0.9018,
+      "step": 8130
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 5.137983172211436e-05,
+      "loss": 0.8998,
+      "step": 8135
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 5.135258515706262e-05,
+      "loss": 0.8722,
+      "step": 8140
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 5.132532590874148e-05,
+      "loss": 0.943,
+      "step": 8145
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 5.1298053998293524e-05,
+      "loss": 0.8835,
+      "step": 8150
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 5.1270769446871184e-05,
+      "loss": 0.9622,
+      "step": 8155
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 5.124347227563671e-05,
+      "loss": 0.8685,
+      "step": 8160
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 5.121616250576213e-05,
+      "loss": 0.9002,
+      "step": 8165
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 5.1188840158429235e-05,
+      "loss": 0.8634,
+      "step": 8170
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 5.116150525482957e-05,
+      "loss": 0.8637,
+      "step": 8175
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 5.113415781616446e-05,
+      "loss": 0.9006,
+      "step": 8180
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 5.1106797863644905e-05,
+      "loss": 0.9038,
+      "step": 8185
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 5.10794254184916e-05,
+      "loss": 0.8638,
+      "step": 8190
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 5.1052040501934995e-05,
+      "loss": 0.8891,
+      "step": 8195
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 5.1024643135215146e-05,
+      "loss": 0.8866,
+      "step": 8200
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 5.099723333958182e-05,
+      "loss": 0.8798,
+      "step": 8205
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 5.096981113629436e-05,
+      "loss": 0.8933,
+      "step": 8210
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 5.09423765466218e-05,
+      "loss": 0.8373,
+      "step": 8215
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 5.091492959184273e-05,
+      "loss": 0.9327,
+      "step": 8220
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 5.088747029324537e-05,
+      "loss": 0.8669,
+      "step": 8225
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 5.085999867212749e-05,
+      "loss": 0.9072,
+      "step": 8230
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 5.083251474979642e-05,
+      "loss": 0.8496,
+      "step": 8235
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 5.080501854756903e-05,
+      "loss": 0.9297,
+      "step": 8240
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 5.077751008677173e-05,
+      "loss": 0.9229,
+      "step": 8245
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 5.074998938874041e-05,
+      "loss": 0.8711,
+      "step": 8250
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 5.072245647482046e-05,
+      "loss": 0.9196,
+      "step": 8255
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 5.0694911366366786e-05,
+      "loss": 0.9173,
+      "step": 8260
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 5.06673540847437e-05,
+      "loss": 0.9033,
+      "step": 8265
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 5.063978465132496e-05,
+      "loss": 0.8882,
+      "step": 8270
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 5.0612203087493785e-05,
+      "loss": 0.8697,
+      "step": 8275
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 5.058460941464276e-05,
+      "loss": 0.8782,
+      "step": 8280
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 5.055700365417391e-05,
+      "loss": 0.8893,
+      "step": 8285
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 5.052938582749858e-05,
+      "loss": 0.9019,
+      "step": 8290
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.0501755956037524e-05,
+      "loss": 0.8657,
+      "step": 8295
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.0474114061220794e-05,
+      "loss": 0.8621,
+      "step": 8300
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.044646016448782e-05,
+      "loss": 0.9236,
+      "step": 8305
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.041879428728728e-05,
+      "loss": 0.8894,
+      "step": 8310
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.039111645107718e-05,
+      "loss": 0.88,
+      "step": 8315
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.0363426677324805e-05,
+      "loss": 0.8776,
+      "step": 8320
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.0335724987506686e-05,
+      "loss": 0.8987,
+      "step": 8325
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.0308011403108594e-05,
+      "loss": 0.8493,
+      "step": 8330
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.0280285945625535e-05,
+      "loss": 0.8899,
+      "step": 8335
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.025254863656173e-05,
+      "loss": 0.871,
+      "step": 8340
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.022479949743057e-05,
+      "loss": 0.9142,
+      "step": 8345
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.019703854975464e-05,
+      "loss": 0.8807,
+      "step": 8350
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.016926581506567e-05,
+      "loss": 0.9072,
+      "step": 8355
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.0141481314904545e-05,
+      "loss": 0.8525,
+      "step": 8360
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.0113685070821285e-05,
+      "loss": 0.9011,
+      "step": 8365
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.008587710437498e-05,
+      "loss": 0.9213,
+      "step": 8370
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.005805743713386e-05,
+      "loss": 0.9034,
+      "step": 8375
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.0030226090675174e-05,
+      "loss": 0.8884,
+      "step": 8380
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.0002383086585295e-05,
+      "loss": 0.8674,
+      "step": 8385
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.9974528446459605e-05,
+      "loss": 0.8865,
+      "step": 8390
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.994666219190247e-05,
+      "loss": 0.8507,
+      "step": 8395
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.9918784344527355e-05,
+      "loss": 0.9086,
+      "step": 8400
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.9890894925956635e-05,
+      "loss": 0.9114,
+      "step": 8405
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.9862993957821714e-05,
+      "loss": 0.8835,
+      "step": 8410
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.9835081461762906e-05,
+      "loss": 0.8812,
+      "step": 8415
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.9807157459429514e-05,
+      "loss": 0.8674,
+      "step": 8420
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.977922197247973e-05,
+      "loss": 0.8719,
+      "step": 8425
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.975127502258067e-05,
+      "loss": 0.9026,
+      "step": 8430
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.9723316631408346e-05,
+      "loss": 0.8867,
+      "step": 8435
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.969534682064762e-05,
+      "loss": 0.8844,
+      "step": 8440
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.9667365611992244e-05,
+      "loss": 0.8705,
+      "step": 8445
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.9639373027144774e-05,
+      "loss": 0.8836,
+      "step": 8450
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.961136908781663e-05,
+      "loss": 0.9074,
+      "step": 8455
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.9583353815728e-05,
+      "loss": 0.8938,
+      "step": 8460
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.9555327232607895e-05,
+      "loss": 0.883,
+      "step": 8465
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.9527289360194065e-05,
+      "loss": 0.884,
+      "step": 8470
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.9499240220233043e-05,
+      "loss": 0.914,
+      "step": 8475
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.94711798344801e-05,
+      "loss": 0.8967,
+      "step": 8480
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.944310822469921e-05,
+      "loss": 0.8766,
+      "step": 8485
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.941502541266307e-05,
+      "loss": 0.8748,
+      "step": 8490
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.938693142015305e-05,
+      "loss": 0.8903,
+      "step": 8495
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.935882626895922e-05,
+      "loss": 0.8658,
+      "step": 8500
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.9330709980880284e-05,
+      "loss": 0.8619,
+      "step": 8505
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.930258257772358e-05,
+      "loss": 0.8788,
+      "step": 8510
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.927444408130506e-05,
+      "loss": 0.8679,
+      "step": 8515
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.924629451344931e-05,
+      "loss": 0.8857,
+      "step": 8520
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.921813389598949e-05,
+      "loss": 0.8898,
+      "step": 8525
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.9189962250767314e-05,
+      "loss": 0.8368,
+      "step": 8530
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.916177959963307e-05,
+      "loss": 0.9066,
+      "step": 8535
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.913358596444557e-05,
+      "loss": 0.8935,
+      "step": 8540
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.910538136707215e-05,
+      "loss": 0.8586,
+      "step": 8545
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.9077165829388664e-05,
+      "loss": 0.8876,
+      "step": 8550
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.904893937327941e-05,
+      "loss": 0.877,
+      "step": 8555
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.90207020206372e-05,
+      "loss": 0.8963,
+      "step": 8560
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.8992453793363266e-05,
+      "loss": 0.8757,
+      "step": 8565
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.896419471336729e-05,
+      "loss": 0.8513,
+      "step": 8570
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.893592480256737e-05,
+      "loss": 0.8755,
+      "step": 8575
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.890764408289001e-05,
+      "loss": 0.8575,
+      "step": 8580
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.887935257627005e-05,
+      "loss": 0.9229,
+      "step": 8585
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.88510503046508e-05,
+      "loss": 0.9283,
+      "step": 8590
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.882273728998381e-05,
+      "loss": 0.8899,
+      "step": 8595
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.879441355422901e-05,
+      "loss": 0.9089,
+      "step": 8600
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.876607911935466e-05,
+      "loss": 0.9093,
+      "step": 8605
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.8737734007337296e-05,
+      "loss": 0.9023,
+      "step": 8610
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.8709378240161736e-05,
+      "loss": 0.8547,
+      "step": 8615
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.8681011839821065e-05,
+      "loss": 0.8603,
+      "step": 8620
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.865263482831664e-05,
+      "loss": 0.8526,
+      "step": 8625
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.8624247227658004e-05,
+      "loss": 0.8532,
+      "step": 8630
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.859584905986294e-05,
+      "loss": 0.838,
+      "step": 8635
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.8567440346957416e-05,
+      "loss": 0.8697,
+      "step": 8640
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.8539021110975595e-05,
+      "loss": 0.8708,
+      "step": 8645
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.851059137395979e-05,
+      "loss": 0.9122,
+      "step": 8650
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.848215115796046e-05,
+      "loss": 0.881,
+      "step": 8655
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.8453700485036185e-05,
+      "loss": 0.8782,
+      "step": 8660
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.842523937725367e-05,
+      "loss": 0.8608,
+      "step": 8665
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.8396767856687714e-05,
+      "loss": 0.8459,
+      "step": 8670
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.836828594542115e-05,
+      "loss": 0.864,
+      "step": 8675
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.833979366554495e-05,
+      "loss": 0.9295,
+      "step": 8680
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.831129103915804e-05,
+      "loss": 0.8693,
+      "step": 8685
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.8282778088367455e-05,
+      "loss": 0.8812,
+      "step": 8690
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.825425483528817e-05,
+      "loss": 0.8669,
+      "step": 8695
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.8225721302043166e-05,
+      "loss": 0.8315,
+      "step": 8700
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.8197177510763446e-05,
+      "loss": 0.9162,
+      "step": 8705
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.8168623483587896e-05,
+      "loss": 0.9012,
+      "step": 8710
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.814005924266339e-05,
+      "loss": 0.9168,
+      "step": 8715
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.811148481014471e-05,
+      "loss": 0.892,
+      "step": 8720
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.808290020819453e-05,
+      "loss": 0.8844,
+      "step": 8725
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.8054305458983456e-05,
+      "loss": 0.8518,
+      "step": 8730
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.80257005846899e-05,
+      "loss": 0.8794,
+      "step": 8735
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.799708560750018e-05,
+      "loss": 0.9048,
+      "step": 8740
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.7968460549608424e-05,
+      "loss": 0.9106,
+      "step": 8745
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.793982543321659e-05,
+      "loss": 0.8977,
+      "step": 8750
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.7911180280534414e-05,
+      "loss": 0.8505,
+      "step": 8755
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.788252511377946e-05,
+      "loss": 0.8829,
+      "step": 8760
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.7853859955177005e-05,
+      "loss": 0.9117,
+      "step": 8765
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.782518482696013e-05,
+      "loss": 0.876,
+      "step": 8770
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.779649975136961e-05,
+      "loss": 0.8802,
+      "step": 8775
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.7767804750653945e-05,
+      "loss": 0.9104,
+      "step": 8780
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.7739099847069345e-05,
+      "loss": 0.8834,
+      "step": 8785
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.7710385062879696e-05,
+      "loss": 0.8863,
+      "step": 8790
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.768166042035653e-05,
+      "loss": 0.9074,
+      "step": 8795
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.765292594177904e-05,
+      "loss": 0.8495,
+      "step": 8800
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.7624181649434064e-05,
+      "loss": 0.8607,
+      "step": 8805
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.759542756561602e-05,
+      "loss": 0.9142,
+      "step": 8810
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.756666371262692e-05,
+      "loss": 0.8482,
+      "step": 8815
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.75378901127764e-05,
+      "loss": 0.9211,
+      "step": 8820
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.75091067883816e-05,
+      "loss": 0.8933,
+      "step": 8825
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.7480313761767245e-05,
+      "loss": 0.8876,
+      "step": 8830
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.7451511055265554e-05,
+      "loss": 0.8695,
+      "step": 8835
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.742269869121626e-05,
+      "loss": 0.8988,
+      "step": 8840
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.73938766919666e-05,
+      "loss": 0.8893,
+      "step": 8845
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.736504507987128e-05,
+      "loss": 0.8989,
+      "step": 8850
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.733620387729247e-05,
+      "loss": 0.8623,
+      "step": 8855
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.730735310659973e-05,
+      "loss": 0.8973,
+      "step": 8860
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.727849279017012e-05,
+      "loss": 0.8949,
+      "step": 8865
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.724962295038803e-05,
+      "loss": 0.8747,
+      "step": 8870
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.7220743609645294e-05,
+      "loss": 0.9145,
+      "step": 8875
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.719185479034107e-05,
+      "loss": 0.9004,
+      "step": 8880
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.71629565148819e-05,
+      "loss": 0.8507,
+      "step": 8885
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.7134048805681643e-05,
+      "loss": 0.8449,
+      "step": 8890
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.710513168516147e-05,
+      "loss": 0.8632,
+      "step": 8895
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.7076205175749886e-05,
+      "loss": 0.8622,
+      "step": 8900
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.704726929988262e-05,
+      "loss": 0.9268,
+      "step": 8905
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.701832408000273e-05,
+      "loss": 0.8562,
+      "step": 8910
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.698936953856048e-05,
+      "loss": 0.8633,
+      "step": 8915
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.6960405698013355e-05,
+      "loss": 0.8388,
+      "step": 8920
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.6931432580826094e-05,
+      "loss": 0.8745,
+      "step": 8925
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.690245020947061e-05,
+      "loss": 0.9127,
+      "step": 8930
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.687345860642599e-05,
+      "loss": 0.9004,
+      "step": 8935
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.684445779417846e-05,
+      "loss": 0.871,
+      "step": 8940
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.681544779522144e-05,
+      "loss": 0.9182,
+      "step": 8945
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.678642863205544e-05,
+      "loss": 0.8663,
+      "step": 8950
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.675740032718809e-05,
+      "loss": 0.8821,
+      "step": 8955
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.6728362903134095e-05,
+      "loss": 0.887,
+      "step": 8960
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.669931638241525e-05,
+      "loss": 0.9065,
+      "step": 8965
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.66702607875604e-05,
+      "loss": 0.8827,
+      "step": 8970
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.664119614110542e-05,
+      "loss": 0.8764,
+      "step": 8975
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.661212246559323e-05,
+      "loss": 0.8373,
+      "step": 8980
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.658303978357372e-05,
+      "loss": 0.9277,
+      "step": 8985
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.6553948117603795e-05,
+      "loss": 0.851,
+      "step": 8990
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.6524847490247295e-05,
+      "loss": 0.8883,
+      "step": 8995
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.6495737924075057e-05,
+      "loss": 0.8999,
+      "step": 9000
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.646661944166479e-05,
+      "loss": 0.8843,
+      "step": 9005
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.643749206560119e-05,
+      "loss": 0.86,
+      "step": 9010
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.640835581847579e-05,
+      "loss": 0.8634,
+      "step": 9015
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.6379210722887034e-05,
+      "loss": 0.8623,
+      "step": 9020
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.635005680144022e-05,
+      "loss": 0.9092,
+      "step": 9025
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.6320894076747496e-05,
+      "loss": 0.9229,
+      "step": 9030
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.629172257142785e-05,
+      "loss": 0.8667,
+      "step": 9035
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.6262542308107036e-05,
+      "loss": 0.8932,
+      "step": 9040
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.623335330941766e-05,
+      "loss": 0.8825,
+      "step": 9045
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.620415559799905e-05,
+      "loss": 0.8524,
+      "step": 9050
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.617494919649733e-05,
+      "loss": 0.922,
+      "step": 9055
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.614573412756535e-05,
+      "loss": 0.8754,
+      "step": 9060
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.6116510413862675e-05,
+      "loss": 0.9038,
+      "step": 9065
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.60872780780556e-05,
+      "loss": 0.8715,
+      "step": 9070
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.6058037142817066e-05,
+      "loss": 0.8828,
+      "step": 9075
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.6028787630826724e-05,
+      "loss": 0.8662,
+      "step": 9080
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.5999529564770844e-05,
+      "loss": 0.8416,
+      "step": 9085
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.5970262967342375e-05,
+      "loss": 0.8802,
+      "step": 9090
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.594098786124084e-05,
+      "loss": 0.848,
+      "step": 9095
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.591170426917238e-05,
+      "loss": 0.8377,
+      "step": 9100
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.5882412213849696e-05,
+      "loss": 0.898,
+      "step": 9105
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.585311171799209e-05,
+      "loss": 0.8702,
+      "step": 9110
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.5823802804325416e-05,
+      "loss": 0.893,
+      "step": 9115
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.579448549558199e-05,
+      "loss": 0.853,
+      "step": 9120
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.576515981450072e-05,
+      "loss": 0.8927,
+      "step": 9125
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.573582578382693e-05,
+      "loss": 0.9077,
+      "step": 9130
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.570648342631251e-05,
+      "loss": 0.8562,
+      "step": 9135
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.5677132764715715e-05,
+      "loss": 0.8355,
+      "step": 9140
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.5647773821801306e-05,
+      "loss": 0.8596,
+      "step": 9145
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.5618406620340435e-05,
+      "loss": 0.8902,
+      "step": 9150
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.558903118311069e-05,
+      "loss": 0.8923,
+      "step": 9155
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.555964753289602e-05,
+      "loss": 0.9284,
+      "step": 9160
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.553025569248674e-05,
+      "loss": 0.8804,
+      "step": 9165
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.550085568467954e-05,
+      "loss": 0.9129,
+      "step": 9170
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.547144753227743e-05,
+      "loss": 0.8835,
+      "step": 9175
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.544203125808975e-05,
+      "loss": 0.8765,
+      "step": 9180
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.5412606884932104e-05,
+      "loss": 0.8948,
+      "step": 9185
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.538317443562642e-05,
+      "loss": 0.8725,
+      "step": 9190
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.535373393300089e-05,
+      "loss": 0.9028,
+      "step": 9195
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.5324285399889894e-05,
+      "loss": 0.8643,
+      "step": 9200
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.529482885913413e-05,
+      "loss": 0.857,
+      "step": 9205
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.526536433358042e-05,
+      "loss": 0.8597,
+      "step": 9210
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.523589184608186e-05,
+      "loss": 0.8924,
+      "step": 9215
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.520641141949763e-05,
+      "loss": 0.8673,
+      "step": 9220
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.5176923076693156e-05,
+      "loss": 0.9003,
+      "step": 9225
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.5147426840539944e-05,
+      "loss": 0.868,
+      "step": 9230
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.511792273391565e-05,
+      "loss": 0.9247,
+      "step": 9235
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.5088410779704034e-05,
+      "loss": 0.934,
+      "step": 9240
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.505889100079492e-05,
+      "loss": 0.8988,
+      "step": 9245
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.502936342008424e-05,
+      "loss": 0.9025,
+      "step": 9250
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.499982806047394e-05,
+      "loss": 0.8408,
+      "step": 9255
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.4970284944872024e-05,
+      "loss": 0.8548,
+      "step": 9260
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.4940734096192485e-05,
+      "loss": 0.8509,
+      "step": 9265
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.491117553735534e-05,
+      "loss": 0.8645,
+      "step": 9270
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.4881609291286586e-05,
+      "loss": 0.838,
+      "step": 9275
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.4852035380918156e-05,
+      "loss": 0.8792,
+      "step": 9280
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.482245382918795e-05,
+      "loss": 0.8846,
+      "step": 9285
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.4792864659039796e-05,
+      "loss": 0.8928,
+      "step": 9290
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.4763267893423426e-05,
+      "loss": 0.871,
+      "step": 9295
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.4733663555294445e-05,
+      "loss": 0.8845,
+      "step": 9300
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.4704051667614364e-05,
+      "loss": 0.9139,
+      "step": 9305
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.4674432253350524e-05,
+      "loss": 0.8583,
+      "step": 9310
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.4644805335476123e-05,
+      "loss": 0.8762,
+      "step": 9315
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.461517093697015e-05,
+      "loss": 0.8473,
+      "step": 9320
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.4585529080817434e-05,
+      "loss": 0.8544,
+      "step": 9325
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.4555879790008555e-05,
+      "loss": 0.9202,
+      "step": 9330
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.452622308753989e-05,
+      "loss": 0.8641,
+      "step": 9335
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.4496558996413545e-05,
+      "loss": 0.9073,
+      "step": 9340
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.4466887539637336e-05,
+      "loss": 0.8749,
+      "step": 9345
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.4437208740224834e-05,
+      "loss": 0.9116,
+      "step": 9350
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.44075226211953e-05,
+      "loss": 0.8907,
+      "step": 9355
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.437782920557363e-05,
+      "loss": 0.8742,
+      "step": 9360
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.434812851639043e-05,
+      "loss": 0.8701,
+      "step": 9365
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.4318420576681905e-05,
+      "loss": 0.8752,
+      "step": 9370
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.428870540948994e-05,
+      "loss": 0.8355,
+      "step": 9375
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.425898303786196e-05,
+      "loss": 0.8699,
+      "step": 9380
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.422925348485101e-05,
+      "loss": 0.8534,
+      "step": 9385
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.4199516773515705e-05,
+      "loss": 0.8555,
+      "step": 9390
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.4169772926920225e-05,
+      "loss": 0.8489,
+      "step": 9395
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.414002196813426e-05,
+      "loss": 0.889,
+      "step": 9400
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.411026392023301e-05,
+      "loss": 0.8982,
+      "step": 9405
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.4080498806297216e-05,
+      "loss": 0.8866,
+      "step": 9410
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.4050726649413045e-05,
+      "loss": 0.8501,
+      "step": 9415
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.402094747267218e-05,
+      "loss": 0.8469,
+      "step": 9420
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.39911612991717e-05,
+      "loss": 0.8528,
+      "step": 9425
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.3961368152014145e-05,
+      "loss": 0.8571,
+      "step": 9430
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.393156805430746e-05,
+      "loss": 0.9167,
+      "step": 9435
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.390176102916495e-05,
+      "loss": 0.9221,
+      "step": 9440
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.3871947099705344e-05,
+      "loss": 0.8675,
+      "step": 9445
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.384212628905267e-05,
+      "loss": 0.8646,
+      "step": 9450
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.381229862033636e-05,
+      "loss": 0.8654,
+      "step": 9455
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.3782464116691085e-05,
+      "loss": 0.8796,
+      "step": 9460
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.375262280125688e-05,
+      "loss": 0.8715,
+      "step": 9465
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.372277469717903e-05,
+      "loss": 0.8698,
+      "step": 9470
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.369291982760812e-05,
+      "loss": 0.9027,
+      "step": 9475
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.3663058215699946e-05,
+      "loss": 0.8932,
+      "step": 9480
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.363318988461554e-05,
+      "loss": 0.8815,
+      "step": 9485
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.360331485752117e-05,
+      "loss": 0.8553,
+      "step": 9490
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.357343315758825e-05,
+      "loss": 0.8679,
+      "step": 9495
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.354354480799345e-05,
+      "loss": 0.9334,
+      "step": 9500
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.35136498319185e-05,
+      "loss": 0.9027,
+      "step": 9505
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.3483748252550344e-05,
+      "loss": 0.9099,
+      "step": 9510
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.3453840093081e-05,
+      "loss": 0.8792,
+      "step": 9515
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.3423925376707635e-05,
+      "loss": 0.8549,
+      "step": 9520
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.3394004126632465e-05,
+      "loss": 0.9129,
+      "step": 9525
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.336407636606278e-05,
+      "loss": 0.8914,
+      "step": 9530
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.333414211821093e-05,
+      "loss": 0.8754,
+      "step": 9535
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.3304201406294295e-05,
+      "loss": 0.8719,
+      "step": 9540
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.3274254253535265e-05,
+      "loss": 0.8836,
+      "step": 9545
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.324430068316121e-05,
+      "loss": 0.8467,
+      "step": 9550
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.321434071840452e-05,
+      "loss": 0.8652,
+      "step": 9555
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.31843743825025e-05,
+      "loss": 0.9038,
+      "step": 9560
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.315440169869742e-05,
+      "loss": 0.8451,
+      "step": 9565
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.3124422690236466e-05,
+      "loss": 0.9137,
+      "step": 9570
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.309443738037173e-05,
+      "loss": 0.8866,
+      "step": 9575
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.306444579236021e-05,
+      "loss": 0.8542,
+      "step": 9580
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.303444794946374e-05,
+      "loss": 0.8559,
+      "step": 9585
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.3004443874949014e-05,
+      "loss": 0.8769,
+      "step": 9590
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.297443359208758e-05,
+      "loss": 0.8331,
+      "step": 9595
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.2944417124155794e-05,
+      "loss": 0.8829,
+      "step": 9600
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.2914394494434796e-05,
+      "loss": 0.8707,
+      "step": 9605
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.28843657262105e-05,
+      "loss": 0.8576,
+      "step": 9610
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.285433084277362e-05,
+      "loss": 0.8345,
+      "step": 9615
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.282428986741957e-05,
+      "loss": 0.8889,
+      "step": 9620
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.2794242823448513e-05,
+      "loss": 0.8789,
+      "step": 9625
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.27641897341653e-05,
+      "loss": 0.8633,
+      "step": 9630
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.27341306228795e-05,
+      "loss": 0.903,
+      "step": 9635
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.2704065512905325e-05,
+      "loss": 0.9004,
+      "step": 9640
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.267399442756165e-05,
+      "loss": 0.8691,
+      "step": 9645
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.264391739017198e-05,
+      "loss": 0.8915,
+      "step": 9650
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.261383442406446e-05,
+      "loss": 0.8816,
+      "step": 9655
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.2583745552571796e-05,
+      "loss": 0.8617,
+      "step": 9660
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.2553650799031294e-05,
+      "loss": 0.8694,
+      "step": 9665
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.2523550186784836e-05,
+      "loss": 0.8975,
+      "step": 9670
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.2493443739178805e-05,
+      "loss": 0.8334,
+      "step": 9675
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.2463331479564185e-05,
+      "loss": 0.8576,
+      "step": 9680
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.243321343129637e-05,
+      "loss": 0.8614,
+      "step": 9685
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.240308961773531e-05,
+      "loss": 0.8601,
+      "step": 9690
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.2372960062245435e-05,
+      "loss": 0.8898,
+      "step": 9695
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.234282478819558e-05,
+      "loss": 0.8484,
+      "step": 9700
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.2312683818959056e-05,
+      "loss": 0.8648,
+      "step": 9705
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.228253717791355e-05,
+      "loss": 0.8646,
+      "step": 9710
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.225238488844121e-05,
+      "loss": 0.894,
+      "step": 9715
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.22222269739285e-05,
+      "loss": 0.8532,
+      "step": 9720
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.219206345776629e-05,
+      "loss": 0.823,
+      "step": 9725
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.216189436334976e-05,
+      "loss": 0.8959,
+      "step": 9730
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.213171971407847e-05,
+      "loss": 0.894,
+      "step": 9735
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.210153953335622e-05,
+      "loss": 0.8895,
+      "step": 9740
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.207135384459116e-05,
+      "loss": 0.857,
+      "step": 9745
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.204116267119568e-05,
+      "loss": 0.8825,
+      "step": 9750
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.2010966036586424e-05,
+      "loss": 0.8903,
+      "step": 9755
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.19807639641843e-05,
+      "loss": 0.828,
+      "step": 9760
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.1950556477414384e-05,
+      "loss": 0.9497,
+      "step": 9765
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.192034359970601e-05,
+      "loss": 0.8555,
+      "step": 9770
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.1890125354492635e-05,
+      "loss": 0.8867,
+      "step": 9775
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.185990176521193e-05,
+      "loss": 0.8303,
+      "step": 9780
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.182967285530568e-05,
+      "loss": 0.8667,
+      "step": 9785
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.1799438648219795e-05,
+      "loss": 0.8951,
+      "step": 9790
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.176919916740432e-05,
+      "loss": 0.87,
+      "step": 9795
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.1738954436313375e-05,
+      "loss": 0.8799,
+      "step": 9800
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.1708704478405144e-05,
+      "loss": 0.9182,
+      "step": 9805
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.1678449317141866e-05,
+      "loss": 0.8693,
+      "step": 9810
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.164818897598983e-05,
+      "loss": 0.9183,
+      "step": 9815
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.161792347841934e-05,
+      "loss": 0.8867,
+      "step": 9820
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.1587652847904677e-05,
+      "loss": 0.8318,
+      "step": 9825
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.155737710792415e-05,
+      "loss": 0.8395,
+      "step": 9830
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.152709628195997e-05,
+      "loss": 0.9418,
+      "step": 9835
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.149681039349835e-05,
+      "loss": 0.838,
+      "step": 9840
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.1466519466029375e-05,
+      "loss": 0.8881,
+      "step": 9845
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.143622352304709e-05,
+      "loss": 0.8805,
+      "step": 9850
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.1405922588049394e-05,
+      "loss": 0.8409,
+      "step": 9855
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.137561668453809e-05,
+      "loss": 0.8352,
+      "step": 9860
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.1345305836018794e-05,
+      "loss": 0.885,
+      "step": 9865
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.1314990066000976e-05,
+      "loss": 0.8875,
+      "step": 9870
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.128466939799795e-05,
+      "loss": 0.8583,
+      "step": 9875
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.125434385552679e-05,
+      "loss": 0.8502,
+      "step": 9880
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.122401346210837e-05,
+      "loss": 0.8912,
+      "step": 9885
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.119367824126732e-05,
+      "loss": 0.8767,
+      "step": 9890
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.1163338216532016e-05,
+      "loss": 0.8842,
+      "step": 9895
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.113299341143457e-05,
+      "loss": 0.8765,
+      "step": 9900
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.1102643849510755e-05,
+      "loss": 0.8542,
+      "step": 9905
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.107228955430012e-05,
+      "loss": 0.8644,
+      "step": 9910
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.104193054934581e-05,
+      "loss": 0.8905,
+      "step": 9915
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.1011566858194665e-05,
+      "loss": 0.94,
+      "step": 9920
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.0981198504397117e-05,
+      "loss": 0.8693,
+      "step": 9925
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.095082551150726e-05,
+      "loss": 0.9064,
+      "step": 9930
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.0920447903082755e-05,
+      "loss": 0.8332,
+      "step": 9935
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.089006570268487e-05,
+      "loss": 0.8607,
+      "step": 9940
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.085967893387841e-05,
+      "loss": 0.8569,
+      "step": 9945
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.082928762023173e-05,
+      "loss": 0.888,
+      "step": 9950
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.0798891785316726e-05,
+      "loss": 0.8899,
+      "step": 9955
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.076849145270876e-05,
+      "loss": 0.8612,
+      "step": 9960
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.0738086645986754e-05,
+      "loss": 0.8475,
+      "step": 9965
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.070767738873301e-05,
+      "loss": 0.8822,
+      "step": 9970
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.067726370453336e-05,
+      "loss": 0.8552,
+      "step": 9975
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.064684561697701e-05,
+      "loss": 0.8951,
+      "step": 9980
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.0616423149656636e-05,
+      "loss": 0.8952,
+      "step": 9985
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.058599632616827e-05,
+      "loss": 0.8311,
+      "step": 9990
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.0555565170111334e-05,
+      "loss": 0.9078,
+      "step": 9995
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.0525129705088625e-05,
+      "loss": 0.8678,
+      "step": 10000
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.0494689954706265e-05,
+      "loss": 0.8949,
+      "step": 10005
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.04642459425737e-05,
+      "loss": 0.861,
+      "step": 10010
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.043379769230369e-05,
+      "loss": 0.907,
+      "step": 10015
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.040334522751227e-05,
+      "loss": 0.8689,
+      "step": 10020
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.0372888571818764e-05,
+      "loss": 0.8887,
+      "step": 10025
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.034242774884573e-05,
+      "loss": 0.8715,
+      "step": 10030
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.031196278221897e-05,
+      "loss": 0.8505,
+      "step": 10035
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.028149369556747e-05,
+      "loss": 0.8603,
+      "step": 10040
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.025102051252348e-05,
+      "loss": 0.8431,
+      "step": 10045
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.022054325672233e-05,
+      "loss": 0.8641,
+      "step": 10050
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.019006195180259e-05,
+      "loss": 0.8862,
+      "step": 10055
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.0159576621405923e-05,
+      "loss": 0.8356,
+      "step": 10060
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.012908728917716e-05,
+      "loss": 0.8859,
+      "step": 10065
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.0098593978764186e-05,
+      "loss": 0.8939,
+      "step": 10070
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.006809671381799e-05,
+      "loss": 0.8852,
+      "step": 10075
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.003759551799265e-05,
+      "loss": 0.8635,
+      "step": 10080
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.0007090414945264e-05,
+      "loss": 0.8947,
+      "step": 10085
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.997658142833598e-05,
+      "loss": 0.862,
+      "step": 10090
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.994606858182793e-05,
+      "loss": 0.8956,
+      "step": 10095
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.9915551899087276e-05,
+      "loss": 0.8697,
+      "step": 10100
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.988503140378314e-05,
+      "loss": 0.8215,
+      "step": 10105
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.985450711958758e-05,
+      "loss": 0.8812,
+      "step": 10110
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.9823979070175636e-05,
+      "loss": 0.8752,
+      "step": 10115
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.979344727922523e-05,
+      "loss": 0.8075,
+      "step": 10120
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.9762911770417206e-05,
+      "loss": 0.845,
+      "step": 10125
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.973237256743529e-05,
+      "loss": 0.8543,
+      "step": 10130
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.970182969396606e-05,
+      "loss": 0.8749,
+      "step": 10135
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.9671283173698954e-05,
+      "loss": 0.868,
+      "step": 10140
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.964073303032624e-05,
+      "loss": 0.8644,
+      "step": 10145
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.9610179287542984e-05,
+      "loss": 0.9115,
+      "step": 10150
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.957962196904706e-05,
+      "loss": 0.8865,
+      "step": 10155
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.9549061098539094e-05,
+      "loss": 0.9017,
+      "step": 10160
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.9518496699722486e-05,
+      "loss": 0.894,
+      "step": 10165
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.948792879630337e-05,
+      "loss": 0.8296,
+      "step": 10170
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.9457357411990586e-05,
+      "loss": 0.8763,
+      "step": 10175
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.942678257049568e-05,
+      "loss": 0.8703,
+      "step": 10180
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.93962042955329e-05,
+      "loss": 0.8656,
+      "step": 10185
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.9365622610819095e-05,
+      "loss": 0.9008,
+      "step": 10190
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.9335037540073855e-05,
+      "loss": 0.8753,
+      "step": 10195
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.9304449107019316e-05,
+      "loss": 0.8426,
+      "step": 10200
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.9273857335380255e-05,
+      "loss": 0.9022,
+      "step": 10205
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.924326224888402e-05,
+      "loss": 0.8742,
+      "step": 10210
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.921266387126055e-05,
+      "loss": 0.8351,
+      "step": 10215
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.918206222624234e-05,
+      "loss": 0.8789,
+      "step": 10220
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.9151457337564403e-05,
+      "loss": 0.852,
+      "step": 10225
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.9120849228964255e-05,
+      "loss": 0.8848,
+      "step": 10230
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.909023792418195e-05,
+      "loss": 0.8671,
+      "step": 10235
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.905962344696e-05,
+      "loss": 0.8364,
+      "step": 10240
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.9029005821043356e-05,
+      "loss": 0.859,
+      "step": 10245
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.8998385070179446e-05,
+      "loss": 0.8995,
+      "step": 10250
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.896776121811809e-05,
+      "loss": 0.8809,
+      "step": 10255
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.8937134288611566e-05,
+      "loss": 0.8777,
+      "step": 10260
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.890650430541447e-05,
+      "loss": 0.9028,
+      "step": 10265
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.8875871292283795e-05,
+      "loss": 0.8836,
+      "step": 10270
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.8845235272978927e-05,
+      "loss": 0.9068,
+      "step": 10275
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.881459627126151e-05,
+      "loss": 0.9066,
+      "step": 10280
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.878395431089556e-05,
+      "loss": 0.8962,
+      "step": 10285
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.875330941564735e-05,
+      "loss": 0.8492,
+      "step": 10290
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.872266160928545e-05,
+      "loss": 0.8376,
+      "step": 10295
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.8692010915580676e-05,
+      "loss": 0.8648,
+      "step": 10300
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.86613573583061e-05,
+      "loss": 0.8847,
+      "step": 10305
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.863070096123698e-05,
+      "loss": 0.9081,
+      "step": 10310
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.860004174815084e-05,
+      "loss": 0.8974,
+      "step": 10315
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.856937974282731e-05,
+      "loss": 0.7935,
+      "step": 10320
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.853871496904824e-05,
+      "loss": 0.8843,
+      "step": 10325
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.850804745059762e-05,
+      "loss": 0.8831,
+      "step": 10330
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.8477377211261546e-05,
+      "loss": 0.854,
+      "step": 10335
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.844670427482826e-05,
+      "loss": 0.8872,
+      "step": 10340
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.8416028665088044e-05,
+      "loss": 0.8451,
+      "step": 10345
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.8385350405833316e-05,
+      "loss": 0.8527,
+      "step": 10350
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.835466952085849e-05,
+      "loss": 0.8691,
+      "step": 10355
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.832398603396007e-05,
+      "loss": 0.8401,
+      "step": 10360
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.829329996893653e-05,
+      "loss": 0.8654,
+      "step": 10365
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.826261134958837e-05,
+      "loss": 0.8647,
+      "step": 10370
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.8231920199718074e-05,
+      "loss": 0.8408,
+      "step": 10375
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.820122654313008e-05,
+      "loss": 0.8512,
+      "step": 10380
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.8170530403630775e-05,
+      "loss": 0.8411,
+      "step": 10385
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.8139831805028456e-05,
+      "loss": 0.8633,
+      "step": 10390
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.810913077113335e-05,
+      "loss": 0.8793,
+      "step": 10395
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.8078427325757554e-05,
+      "loss": 0.8976,
+      "step": 10400
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.804772149271506e-05,
+      "loss": 0.8648,
+      "step": 10405
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.801701329582168e-05,
+      "loss": 0.8803,
+      "step": 10410
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.7986302758895074e-05,
+      "loss": 0.8643,
+      "step": 10415
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.7955589905754736e-05,
+      "loss": 0.885,
+      "step": 10420
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.7924874760221925e-05,
+      "loss": 0.9237,
+      "step": 10425
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.7894157346119695e-05,
+      "loss": 0.8613,
+      "step": 10430
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.7863437687272845e-05,
+      "loss": 0.8653,
+      "step": 10435
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.783271580750795e-05,
+      "loss": 0.8688,
+      "step": 10440
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.780199173065327e-05,
+      "loss": 0.8422,
+      "step": 10445
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.777126548053878e-05,
+      "loss": 0.8716,
+      "step": 10450
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.774053708099615e-05,
+      "loss": 0.8521,
+      "step": 10455
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.7709806555858696e-05,
+      "loss": 0.8592,
+      "step": 10460
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.7679073928961426e-05,
+      "loss": 0.8337,
+      "step": 10465
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.764833922414093e-05,
+      "loss": 0.9027,
+      "step": 10470
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.761760246523543e-05,
+      "loss": 0.852,
+      "step": 10475
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.7586863676084734e-05,
+      "loss": 0.8617,
+      "step": 10480
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.755612288053026e-05,
+      "loss": 0.8599,
+      "step": 10485
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.752538010241492e-05,
+      "loss": 0.8314,
+      "step": 10490
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.749463536558321e-05,
+      "loss": 0.8451,
+      "step": 10495
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.7463888693881136e-05,
+      "loss": 0.8922,
+      "step": 10500
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.743314011115619e-05,
+      "loss": 0.8742,
+      "step": 10505
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.740238964125738e-05,
+      "loss": 0.8702,
+      "step": 10510
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.7371637308035126e-05,
+      "loss": 0.885,
+      "step": 10515
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.734088313534134e-05,
+      "loss": 0.8864,
+      "step": 10520
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.731012714702932e-05,
+      "loss": 0.8413,
+      "step": 10525
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.727936936695383e-05,
+      "loss": 0.8673,
+      "step": 10530
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.7248609818970975e-05,
+      "loss": 0.8529,
+      "step": 10535
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.721784852693823e-05,
+      "loss": 0.8984,
+      "step": 10540
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.718708551471445e-05,
+      "loss": 0.8664,
+      "step": 10545
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.7156320806159816e-05,
+      "loss": 0.8634,
+      "step": 10550
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.7125554425135814e-05,
+      "loss": 0.8694,
+      "step": 10555
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.7094786395505234e-05,
+      "loss": 0.8847,
+      "step": 10560
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.706401674113214e-05,
+      "loss": 0.8676,
+      "step": 10565
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.7033245485881876e-05,
+      "loss": 0.8967,
+      "step": 10570
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.7002472653620984e-05,
+      "loss": 0.8738,
+      "step": 10575
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.6971698268217284e-05,
+      "loss": 0.858,
+      "step": 10580
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.6940922353539766e-05,
+      "loss": 0.8954,
+      "step": 10585
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.691014493345862e-05,
+      "loss": 0.8375,
+      "step": 10590
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.687936603184518e-05,
+      "loss": 0.8578,
+      "step": 10595
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.6848585672571953e-05,
+      "loss": 0.871,
+      "step": 10600
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.681780387951258e-05,
+      "loss": 0.8741,
+      "step": 10605
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.6787020676541797e-05,
+      "loss": 0.9203,
+      "step": 10610
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.675623608753545e-05,
+      "loss": 0.8868,
+      "step": 10615
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.672545013637043e-05,
+      "loss": 0.8862,
+      "step": 10620
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.669466284692474e-05,
+      "loss": 0.831,
+      "step": 10625
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.666387424307735e-05,
+      "loss": 0.8204,
+      "step": 10630
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.6633084348708316e-05,
+      "loss": 0.8524,
+      "step": 10635
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.6602293187698626e-05,
+      "loss": 0.8424,
+      "step": 10640
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.6571500783930335e-05,
+      "loss": 0.8481,
+      "step": 10645
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.6540707161286396e-05,
+      "loss": 0.8722,
+      "step": 10650
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.6509912343650706e-05,
+      "loss": 0.8812,
+      "step": 10655
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.647911635490815e-05,
+      "loss": 0.8764,
+      "step": 10660
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.644831921894446e-05,
+      "loss": 0.855,
+      "step": 10665
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.6417520959646285e-05,
+      "loss": 0.853,
+      "step": 10670
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.638672160090113e-05,
+      "loss": 0.8337,
+      "step": 10675
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.6355921166597374e-05,
+      "loss": 0.8774,
+      "step": 10680
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.6325119680624205e-05,
+      "loss": 0.836,
+      "step": 10685
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.629431716687166e-05,
+      "loss": 0.8563,
+      "step": 10690
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.6263513649230545e-05,
+      "loss": 0.8756,
+      "step": 10695
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.6232709151592444e-05,
+      "loss": 0.8763,
+      "step": 10700
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.620190369784973e-05,
+      "loss": 0.8824,
+      "step": 10705
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.617109731189548e-05,
+      "loss": 0.8671,
+      "step": 10710
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.614029001762352e-05,
+      "loss": 0.8568,
+      "step": 10715
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.6109481838928365e-05,
+      "loss": 0.8685,
+      "step": 10720
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.607867279970523e-05,
+      "loss": 0.8381,
+      "step": 10725
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.6047862923850006e-05,
+      "loss": 0.8537,
+      "step": 10730
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.601705223525918e-05,
+      "loss": 0.866,
+      "step": 10735
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.5986240757829936e-05,
+      "loss": 0.8608,
+      "step": 10740
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.595542851546003e-05,
+      "loss": 0.8551,
+      "step": 10745
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.592461553204785e-05,
+      "loss": 0.8751,
+      "step": 10750
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.5893801831492286e-05,
+      "loss": 0.8852,
+      "step": 10755
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.5862987437692854e-05,
+      "loss": 0.8642,
+      "step": 10760
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.5832172374549576e-05,
+      "loss": 0.8829,
+      "step": 10765
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.580135666596301e-05,
+      "loss": 0.86,
+      "step": 10770
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.57705403358342e-05,
+      "loss": 0.8534,
+      "step": 10775
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.573972340806467e-05,
+      "loss": 0.8484,
+      "step": 10780
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.5708905906556426e-05,
+      "loss": 0.8567,
+      "step": 10785
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.5678087855211895e-05,
+      "loss": 0.8456,
+      "step": 10790
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.5647269277933955e-05,
+      "loss": 0.8434,
+      "step": 10795
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.561645019862586e-05,
+      "loss": 0.8937,
+      "step": 10800
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.55856306411913e-05,
+      "loss": 0.8311,
+      "step": 10805
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.555481062953429e-05,
+      "loss": 0.866,
+      "step": 10810
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.552399018755922e-05,
+      "loss": 0.8349,
+      "step": 10815
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.549316933917082e-05,
+      "loss": 0.8512,
+      "step": 10820
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.54623481082741e-05,
+      "loss": 0.885,
+      "step": 10825
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.5431526518774435e-05,
+      "loss": 0.8834,
+      "step": 10830
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.540070459457738e-05,
+      "loss": 0.8294,
+      "step": 10835
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.5369882359588856e-05,
+      "loss": 0.8803,
+      "step": 10840
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.533905983771493e-05,
+      "loss": 0.8769,
+      "step": 10845
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.5308237052861974e-05,
+      "loss": 0.8698,
+      "step": 10850
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.5277414028936496e-05,
+      "loss": 0.8314,
+      "step": 10855
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.524659078984523e-05,
+      "loss": 0.8413,
+      "step": 10860
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.521576735949507e-05,
+      "loss": 0.8664,
+      "step": 10865
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.518494376179306e-05,
+      "loss": 0.8552,
+      "step": 10870
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.515412002064635e-05,
+      "loss": 0.8384,
+      "step": 10875
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.512329615996221e-05,
+      "loss": 0.8055,
+      "step": 10880
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.5092472203648045e-05,
+      "loss": 0.8702,
+      "step": 10885
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.506164817561128e-05,
+      "loss": 0.8227,
+      "step": 10890
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.503082409975943e-05,
+      "loss": 0.828,
+      "step": 10895
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.5e-05,
+      "loss": 0.8562,
+      "step": 10900
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.4969175900240584e-05,
+      "loss": 0.8468,
+      "step": 10905
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.493835182438871e-05,
+      "loss": 0.8915,
+      "step": 10910
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.490752779635195e-05,
+      "loss": 0.85,
+      "step": 10915
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.4876703840037776e-05,
+      "loss": 0.8712,
+      "step": 10920
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.484587997935365e-05,
+      "loss": 0.8758,
+      "step": 10925
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.481505623820695e-05,
+      "loss": 0.8244,
+      "step": 10930
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.4784232640504914e-05,
+      "loss": 0.8685,
+      "step": 10935
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.4753409210154766e-05,
+      "loss": 0.8056,
+      "step": 10940
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.4722585971063505e-05,
+      "loss": 0.8146,
+      "step": 10945
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.4691762947138027e-05,
+      "loss": 0.9343,
+      "step": 10950
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.466094016228507e-05,
+      "loss": 0.824,
+      "step": 10955
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.463011764041114e-05,
+      "loss": 0.883,
+      "step": 10960
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.4599295405422615e-05,
+      "loss": 0.8461,
+      "step": 10965
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.456847348122558e-05,
+      "loss": 0.8779,
+      "step": 10970
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.4537651891725894e-05,
+      "loss": 0.8363,
+      "step": 10975
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.450683066082918e-05,
+      "loss": 0.8429,
+      "step": 10980
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.447600981244078e-05,
+      "loss": 0.8626,
+      "step": 10985
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.4445189370465706e-05,
+      "loss": 0.8444,
+      "step": 10990
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.44143693588087e-05,
+      "loss": 0.8487,
+      "step": 10995
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.438354980137413e-05,
+      "loss": 0.8498,
+      "step": 11000
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.4352730722066046e-05,
+      "loss": 0.8551,
+      "step": 11005
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.432191214478811e-05,
+      "loss": 0.8397,
+      "step": 11010
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.429109409344357e-05,
+      "loss": 0.8515,
+      "step": 11015
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.426027659193533e-05,
+      "loss": 0.8446,
+      "step": 11020
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.422945966416581e-05,
+      "loss": 0.8386,
+      "step": 11025
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.419864333403699e-05,
+      "loss": 0.8196,
+      "step": 11030
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.416782762545042e-05,
+      "loss": 0.8566,
+      "step": 11035
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.413701256230714e-05,
+      "loss": 0.8479,
+      "step": 11040
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.4106198168507715e-05,
+      "loss": 0.8433,
+      "step": 11045
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.407538446795215e-05,
+      "loss": 0.8449,
+      "step": 11050
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.4044571484539956e-05,
+      "loss": 0.8835,
+      "step": 11055
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.401375924217006e-05,
+      "loss": 0.8621,
+      "step": 11060
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.398294776474082e-05,
+      "loss": 0.8502,
+      "step": 11065
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.395213707615e-05,
+      "loss": 0.8421,
+      "step": 11070
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.392132720029476e-05,
+      "loss": 0.8426,
+      "step": 11075
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.389051816107162e-05,
+      "loss": 0.8912,
+      "step": 11080
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.385970998237648e-05,
+      "loss": 0.8843,
+      "step": 11085
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.382890268810453e-05,
+      "loss": 0.8349,
+      "step": 11090
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.3798096302150266e-05,
+      "loss": 0.8179,
+      "step": 11095
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.376729084840756e-05,
+      "loss": 0.8786,
+      "step": 11100
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.373648635076946e-05,
+      "loss": 0.829,
+      "step": 11105
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.3705682833128335e-05,
+      "loss": 0.8688,
+      "step": 11110
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.3674880319375796e-05,
+      "loss": 0.8305,
+      "step": 11115
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.364407883340263e-05,
+      "loss": 0.8618,
+      "step": 11120
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.361327839909887e-05,
+      "loss": 0.841,
+      "step": 11125
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.358247904035372e-05,
+      "loss": 0.8873,
+      "step": 11130
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.3551680781055536e-05,
+      "loss": 0.909,
+      "step": 11135
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.352088364509185e-05,
+      "loss": 0.8675,
+      "step": 11140
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.349008765634929e-05,
+      "loss": 0.8479,
+      "step": 11145
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.3459292838713605e-05,
+      "loss": 0.8631,
+      "step": 11150
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.342849921606966e-05,
+      "loss": 0.8816,
+      "step": 11155
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.339770681230136e-05,
+      "loss": 0.8744,
+      "step": 11160
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.336691565129169e-05,
+      "loss": 0.8057,
+      "step": 11165
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.333612575692265e-05,
+      "loss": 0.8719,
+      "step": 11170
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.3305337153075254e-05,
+      "loss": 0.8454,
+      "step": 11175
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.3274549863629564e-05,
+      "loss": 0.8554,
+      "step": 11180
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.3243763912464556e-05,
+      "loss": 0.8484,
+      "step": 11185
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.32129793234582e-05,
+      "loss": 0.8786,
+      "step": 11190
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.318219612048742e-05,
+      "loss": 0.8489,
+      "step": 11195
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.315141432742804e-05,
+      "loss": 0.8331,
+      "step": 11200
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.312063396815482e-05,
+      "loss": 0.8523,
+      "step": 11205
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.308985506654139e-05,
+      "loss": 0.8525,
+      "step": 11210
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.305907764646023e-05,
+      "loss": 0.9026,
+      "step": 11215
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.302830173178271e-05,
+      "loss": 0.8568,
+      "step": 11220
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.299752734637902e-05,
+      "loss": 0.8436,
+      "step": 11225
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.2966754514118125e-05,
+      "loss": 0.8756,
+      "step": 11230
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.2935983258867856e-05,
+      "loss": 0.8548,
+      "step": 11235
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.290521360449476e-05,
+      "loss": 0.8658,
+      "step": 11240
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.287444557486418e-05,
+      "loss": 0.8832,
+      "step": 11245
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.2843679193840184e-05,
+      "loss": 0.8444,
+      "step": 11250
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.2812914485285535e-05,
+      "loss": 0.8856,
+      "step": 11255
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.278215147306177e-05,
+      "loss": 0.904,
+      "step": 11260
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.275139018102903e-05,
+      "loss": 0.8454,
+      "step": 11265
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.272063063304616e-05,
+      "loss": 0.8686,
+      "step": 11270
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.268987285297067e-05,
+      "loss": 0.8531,
+      "step": 11275
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.2659116864658664e-05,
+      "loss": 0.8201,
+      "step": 11280
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.262836269196488e-05,
+      "loss": 0.8253,
+      "step": 11285
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.259761035874263e-05,
+      "loss": 0.8503,
+      "step": 11290
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.2566859888843804e-05,
+      "loss": 0.8593,
+      "step": 11295
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.2536111306118864e-05,
+      "loss": 0.8527,
+      "step": 11300
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.250536463441679e-05,
+      "loss": 0.9123,
+      "step": 11305
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.2474619897585074e-05,
+      "loss": 0.8609,
+      "step": 11310
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.244387711946974e-05,
+      "loss": 0.8608,
+      "step": 11315
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.2413136323915247e-05,
+      "loss": 0.8873,
+      "step": 11320
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.238239753476456e-05,
+      "loss": 0.8345,
+      "step": 11325
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.2351660775859074e-05,
+      "loss": 0.9119,
+      "step": 11330
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.232092607103856e-05,
+      "loss": 0.9135,
+      "step": 11335
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.22901934441413e-05,
+      "loss": 0.8549,
+      "step": 11340
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.225946291900386e-05,
+      "loss": 0.8272,
+      "step": 11345
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.222873451946122e-05,
+      "loss": 0.8014,
+      "step": 11350
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.2198008269346734e-05,
+      "loss": 0.8421,
+      "step": 11355
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.216728419249205e-05,
+      "loss": 0.8627,
+      "step": 11360
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.213656231272715e-05,
+      "loss": 0.8098,
+      "step": 11365
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.210584265388031e-05,
+      "loss": 0.8516,
+      "step": 11370
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.2075125239778075e-05,
+      "loss": 0.8157,
+      "step": 11375
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.204441009424526e-05,
+      "loss": 0.8903,
+      "step": 11380
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.201369724110492e-05,
+      "loss": 0.8514,
+      "step": 11385
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.198298670417832e-05,
+      "loss": 0.829,
+      "step": 11390
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.195227850728494e-05,
+      "loss": 0.8494,
+      "step": 11395
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.1921572674242434e-05,
+      "loss": 0.8441,
+      "step": 11400
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.189086922886665e-05,
+      "loss": 0.8277,
+      "step": 11405
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.186016819497155e-05,
+      "loss": 0.8779,
+      "step": 11410
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.182946959636922e-05,
+      "loss": 0.8601,
+      "step": 11415
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.179877345686992e-05,
+      "loss": 0.8724,
+      "step": 11420
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.176807980028191e-05,
+      "loss": 0.8812,
+      "step": 11425
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.173738865041163e-05,
+      "loss": 0.8545,
+      "step": 11430
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.1706700031063476e-05,
+      "loss": 0.8643,
+      "step": 11435
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.167601396603993e-05,
+      "loss": 0.8677,
+      "step": 11440
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.164533047914151e-05,
+      "loss": 0.8192,
+      "step": 11445
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.161464959416669e-05,
+      "loss": 0.8492,
+      "step": 11450
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.158397133491195e-05,
+      "loss": 0.8461,
+      "step": 11455
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.155329572517175e-05,
+      "loss": 0.8584,
+      "step": 11460
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.152262278873845e-05,
+      "loss": 0.8479,
+      "step": 11465
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.149195254940238e-05,
+      "loss": 0.8491,
+      "step": 11470
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.1461285030951765e-05,
+      "loss": 0.8837,
+      "step": 11475
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.1430620257172686e-05,
+      "loss": 0.9047,
+      "step": 11480
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.139995825184916e-05,
+      "loss": 0.8661,
+      "step": 11485
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.136929903876302e-05,
+      "loss": 0.8637,
+      "step": 11490
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.13386426416939e-05,
+      "loss": 0.8444,
+      "step": 11495
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.130798908441933e-05,
+      "loss": 0.8175,
+      "step": 11500
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.1277338390714544e-05,
+      "loss": 0.8203,
+      "step": 11505
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.124669058435265e-05,
+      "loss": 0.8286,
+      "step": 11510
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.121604568910444e-05,
+      "loss": 0.8401,
+      "step": 11515
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.118540372873848e-05,
+      "loss": 0.8418,
+      "step": 11520
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.1154764727021074e-05,
+      "loss": 0.8357,
+      "step": 11525
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.11241287077162e-05,
+      "loss": 0.8627,
+      "step": 11530
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.109349569458553e-05,
+      "loss": 0.8624,
+      "step": 11535
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.106286571138844e-05,
+      "loss": 0.8582,
+      "step": 11540
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.10322387818819e-05,
+      "loss": 0.8699,
+      "step": 11545
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.1001614929820555e-05,
+      "loss": 0.8509,
+      "step": 11550
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.097099417895665e-05,
+      "loss": 0.8633,
+      "step": 11555
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.0940376553039995e-05,
+      "loss": 0.8122,
+      "step": 11560
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.0909762075818044e-05,
+      "loss": 0.8295,
+      "step": 11565
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.0879150771035746e-05,
+      "loss": 0.8451,
+      "step": 11570
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.08485426624356e-05,
+      "loss": 0.8599,
+      "step": 11575
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.081793777375766e-05,
+      "loss": 0.8682,
+      "step": 11580
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.078733612873944e-05,
+      "loss": 0.8816,
+      "step": 11585
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.075673775111598e-05,
+      "loss": 0.8828,
+      "step": 11590
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.072614266461975e-05,
+      "loss": 0.8292,
+      "step": 11595
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.069555089298068e-05,
+      "loss": 0.8633,
+      "step": 11600
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.066496245992614e-05,
+      "loss": 0.8217,
+      "step": 11605
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.06343773891809e-05,
+      "loss": 0.8769,
+      "step": 11610
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.0603795704467104e-05,
+      "loss": 0.829,
+      "step": 11615
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.057321742950431e-05,
+      "loss": 0.8439,
+      "step": 11620
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.054264258800941e-05,
+      "loss": 0.8245,
+      "step": 11625
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.0512071203696627e-05,
+      "loss": 0.8389,
+      "step": 11630
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.0481503300277514e-05,
+      "loss": 0.8455,
+      "step": 11635
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.04509389014609e-05,
+      "loss": 0.8381,
+      "step": 11640
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.0420378030952943e-05,
+      "loss": 0.9062,
+      "step": 11645
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.0389820712457016e-05,
+      "loss": 0.8115,
+      "step": 11650
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.035926696967376e-05,
+      "loss": 0.8371,
+      "step": 11655
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.0328716826301047e-05,
+      "loss": 0.8508,
+      "step": 11660
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.0298170306033938e-05,
+      "loss": 0.8608,
+      "step": 11665
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.026762743256471e-05,
+      "loss": 0.8478,
+      "step": 11670
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.023708822958279e-05,
+      "loss": 0.8589,
+      "step": 11675
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.0206552720774767e-05,
+      "loss": 0.8646,
+      "step": 11680
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.0176020929824362e-05,
+      "loss": 0.8552,
+      "step": 11685
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.014549288041242e-05,
+      "loss": 0.8279,
+      "step": 11690
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.011496859621686e-05,
+      "loss": 0.8774,
+      "step": 11695
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.008444810091272e-05,
+      "loss": 0.8195,
+      "step": 11700
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.0053931418172064e-05,
+      "loss": 0.8402,
+      "step": 11705
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.002341857166402e-05,
+      "loss": 0.8406,
+      "step": 11710
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.999290958505474e-05,
+      "loss": 0.8559,
+      "step": 11715
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.996240448200734e-05,
+      "loss": 0.8744,
+      "step": 11720
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.993190328618201e-05,
+      "loss": 0.8353,
+      "step": 11725
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.9901406021235822e-05,
+      "loss": 0.873,
+      "step": 11730
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.987091271082284e-05,
+      "loss": 0.8567,
+      "step": 11735
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.9840423378594074e-05,
+      "loss": 0.886,
+      "step": 11740
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.9809938048197413e-05,
+      "loss": 0.842,
+      "step": 11745
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.977945674327767e-05,
+      "loss": 0.8327,
+      "step": 11750
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.974897948747653e-05,
+      "loss": 0.8346,
+      "step": 11755
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.9718506304432515e-05,
+      "loss": 0.8523,
+      "step": 11760
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.968803721778103e-05,
+      "loss": 0.8957,
+      "step": 11765
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.9657572251154267e-05,
+      "loss": 0.8268,
+      "step": 11770
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.962711142818123e-05,
+      "loss": 0.8771,
+      "step": 11775
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.9596654772487727e-05,
+      "loss": 0.9198,
+      "step": 11780
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.956620230769631e-05,
+      "loss": 0.8495,
+      "step": 11785
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.9535754057426296e-05,
+      "loss": 0.8614,
+      "step": 11790
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.9505310045293743e-05,
+      "loss": 0.8851,
+      "step": 11795
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.9474870294911366e-05,
+      "loss": 0.8467,
+      "step": 11800
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.9444434829888663e-05,
+      "loss": 0.8364,
+      "step": 11805
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.9414003673831733e-05,
+      "loss": 0.8193,
+      "step": 11810
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.938357685034336e-05,
+      "loss": 0.8402,
+      "step": 11815
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.935315438302299e-05,
+      "loss": 0.8141,
+      "step": 11820
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.9322736295466642e-05,
+      "loss": 0.8459,
+      "step": 11825
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.9292322611266985e-05,
+      "loss": 0.8377,
+      "step": 11830
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.926191335401325e-05,
+      "loss": 0.8282,
+      "step": 11835
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.9231508547291225e-05,
+      "loss": 0.867,
+      "step": 11840
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.9201108214683275e-05,
+      "loss": 0.8084,
+      "step": 11845
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.917071237976826e-05,
+      "loss": 0.8596,
+      "step": 11850
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.914032106612158e-05,
+      "loss": 0.832,
+      "step": 11855
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.9109934297315124e-05,
+      "loss": 0.8553,
+      "step": 11860
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.9079552096917236e-05,
+      "loss": 0.8758,
+      "step": 11865
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.904917448849274e-05,
+      "loss": 0.8258,
+      "step": 11870
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.901880149560289e-05,
+      "loss": 0.8364,
+      "step": 11875
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.8988433141805332e-05,
+      "loss": 0.8649,
+      "step": 11880
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.8958069450654188e-05,
+      "loss": 0.8372,
+      "step": 11885
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.8927710445699865e-05,
+      "loss": 0.8495,
+      "step": 11890
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.8897356150489235e-05,
+      "loss": 0.8522,
+      "step": 11895
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.8867006588565444e-05,
+      "loss": 0.8378,
+      "step": 11900
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.883666178346798e-05,
+      "loss": 0.8813,
+      "step": 11905
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.880632175873268e-05,
+      "loss": 0.8391,
+      "step": 11910
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.877598653789163e-05,
+      "loss": 0.8344,
+      "step": 11915
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.8745656144473203e-05,
+      "loss": 0.8342,
+      "step": 11920
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.8715330602002048e-05,
+      "loss": 0.8396,
+      "step": 11925
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.8685009933999015e-05,
+      "loss": 0.8213,
+      "step": 11930
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.865469416398121e-05,
+      "loss": 0.8642,
+      "step": 11935
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.8624383315461917e-05,
+      "loss": 0.8309,
+      "step": 11940
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.8594077411950596e-05,
+      "loss": 0.8404,
+      "step": 11945
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.856377647695291e-05,
+      "loss": 0.8283,
+      "step": 11950
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.8533480533970633e-05,
+      "loss": 0.8286,
+      "step": 11955
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.850318960650165e-05,
+      "loss": 0.8444,
+      "step": 11960
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.847290371804003e-05,
+      "loss": 0.8384,
+      "step": 11965
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.8442622892075843e-05,
+      "loss": 0.8989,
+      "step": 11970
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.8412347152095314e-05,
+      "loss": 0.8286,
+      "step": 11975
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.838207652158066e-05,
+      "loss": 0.8681,
+      "step": 11980
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.8351811024010164e-05,
+      "loss": 0.8472,
+      "step": 11985
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.8321550682858135e-05,
+      "loss": 0.8445,
+      "step": 11990
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.829129552159486e-05,
+      "loss": 0.7956,
+      "step": 11995
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.826104556368662e-05,
+      "loss": 0.8475,
+      "step": 12000
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.8230800832595675e-05,
+      "loss": 0.8437,
+      "step": 12005
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.8200561351780192e-05,
+      "loss": 0.8348,
+      "step": 12010
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.8170327144694316e-05,
+      "loss": 0.8662,
+      "step": 12015
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.8140098234788066e-05,
+      "loss": 0.7989,
+      "step": 12020
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.8109874645507355e-05,
+      "loss": 0.8776,
+      "step": 12025
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.807965640029399e-05,
+      "loss": 0.8578,
+      "step": 12030
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.8049443522585614e-05,
+      "loss": 0.8181,
+      "step": 12035
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.8019236035815693e-05,
+      "loss": 0.8154,
+      "step": 12040
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.7989033963413573e-05,
+      "loss": 0.7941,
+      "step": 12045
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.795883732880431e-05,
+      "loss": 0.8414,
+      "step": 12050
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.792864615540884e-05,
+      "loss": 0.8455,
+      "step": 12055
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.789846046664378e-05,
+      "loss": 0.7755,
+      "step": 12060
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.7868280285921532e-05,
+      "loss": 0.8169,
+      "step": 12065
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.783810563665024e-05,
+      "loss": 0.83,
+      "step": 12070
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.780793654223372e-05,
+      "loss": 0.8421,
+      "step": 12075
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.7777773026071498e-05,
+      "loss": 0.8242,
+      "step": 12080
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.7747615111558792e-05,
+      "loss": 0.8414,
+      "step": 12085
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.7717462822086437e-05,
+      "loss": 0.8577,
+      "step": 12090
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.768731618104094e-05,
+      "loss": 0.8528,
+      "step": 12095
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.7657175211804422e-05,
+      "loss": 0.8442,
+      "step": 12100
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.762703993775456e-05,
+      "loss": 0.8135,
+      "step": 12105
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.7596910382264682e-05,
+      "loss": 0.8674,
+      "step": 12110
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.7566786568703637e-05,
+      "loss": 0.8427,
+      "step": 12115
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.7536668520435816e-05,
+      "loss": 0.8479,
+      "step": 12120
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.7506556260821185e-05,
+      "loss": 0.8886,
+      "step": 12125
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.7476449813215158e-05,
+      "loss": 0.8442,
+      "step": 12130
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.7446349200968703e-05,
+      "loss": 0.9036,
+      "step": 12135
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.7416254447428205e-05,
+      "loss": 0.8419,
+      "step": 12140
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.738616557593554e-05,
+      "loss": 0.8429,
+      "step": 12145
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.7356082609828016e-05,
+      "loss": 0.8569,
+      "step": 12150
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.7326005572438357e-05,
+      "loss": 0.8606,
+      "step": 12155
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.7295934487094676e-05,
+      "loss": 0.89,
+      "step": 12160
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.72658693771205e-05,
+      "loss": 0.8519,
+      "step": 12165
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.7235810265834692e-05,
+      "loss": 0.8562,
+      "step": 12170
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.7205757176551484e-05,
+      "loss": 0.8506,
+      "step": 12175
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.7175710132580435e-05,
+      "loss": 0.8192,
+      "step": 12180
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.7145669157226373e-05,
+      "loss": 0.8983,
+      "step": 12185
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.7115634273789497e-05,
+      "loss": 0.8119,
+      "step": 12190
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.7085605505565215e-05,
+      "loss": 0.8464,
+      "step": 12195
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.7055582875844207e-05,
+      "loss": 0.8224,
+      "step": 12200
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.702556640791242e-05,
+      "loss": 0.9118,
+      "step": 12205
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.699555612505098e-05,
+      "loss": 0.8527,
+      "step": 12210
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.6965552050536263e-05,
+      "loss": 0.8438,
+      "step": 12215
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.6935554207639787e-05,
+      "loss": 0.8366,
+      "step": 12220
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.6905562619628255e-05,
+      "loss": 0.8061,
+      "step": 12225
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.6875577309763528e-05,
+      "loss": 0.833,
+      "step": 12230
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.684559830130258e-05,
+      "loss": 0.8537,
+      "step": 12235
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.6815625617497494e-05,
+      "loss": 0.8678,
+      "step": 12240
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.6785659281595478e-05,
+      "loss": 0.8581,
+      "step": 12245
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.675569931683878e-05,
+      "loss": 0.8376,
+      "step": 12250
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.6725745746464733e-05,
+      "loss": 0.8534,
+      "step": 12255
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.669579859370571e-05,
+      "loss": 0.8145,
+      "step": 12260
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.6665857881789058e-05,
+      "loss": 0.8093,
+      "step": 12265
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.663592363393722e-05,
+      "loss": 0.8241,
+      "step": 12270
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.6605995873367536e-05,
+      "loss": 0.8631,
+      "step": 12275
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.6576074623292362e-05,
+      "loss": 0.8778,
+      "step": 12280
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.6546159906918994e-05,
+      "loss": 0.8318,
+      "step": 12285
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.6516251747449653e-05,
+      "loss": 0.8143,
+      "step": 12290
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.64863501680815e-05,
+      "loss": 0.8266,
+      "step": 12295
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.6456455192006557e-05,
+      "loss": 0.8208,
+      "step": 12300
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.6426566842411734e-05,
+      "loss": 0.8469,
+      "step": 12305
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.6396685142478836e-05,
+      "loss": 0.8568,
+      "step": 12310
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.636681011538446e-05,
+      "loss": 0.8639,
+      "step": 12315
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.6336941784300055e-05,
+      "loss": 0.8267,
+      "step": 12320
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.6307080172391875e-05,
+      "loss": 0.842,
+      "step": 12325
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.6277225302820958e-05,
+      "loss": 0.8496,
+      "step": 12330
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.6247377198743117e-05,
+      "loss": 0.8349,
+      "step": 12335
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.6217535883308926e-05,
+      "loss": 0.8643,
+      "step": 12340
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.618770137966364e-05,
+      "loss": 0.8538,
+      "step": 12345
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.615787371094732e-05,
+      "loss": 0.871,
+      "step": 12350
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.6128052900294647e-05,
+      "loss": 0.8533,
+      "step": 12355
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.6098238970835043e-05,
+      "loss": 0.8263,
+      "step": 12360
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.6068431945692547e-05,
+      "loss": 0.8615,
+      "step": 12365
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.603863184798585e-05,
+      "loss": 0.8623,
+      "step": 12370
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.60088387008283e-05,
+      "loss": 0.871,
+      "step": 12375
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.5979052527327824e-05,
+      "loss": 0.8633,
+      "step": 12380
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.594927335058695e-05,
+      "loss": 0.8231,
+      "step": 12385
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.591950119370279e-05,
+      "loss": 0.8004,
+      "step": 12390
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.5889736079766986e-05,
+      "loss": 0.8221,
+      "step": 12395
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.585997803186574e-05,
+      "loss": 0.844,
+      "step": 12400
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.5830227073079773e-05,
+      "loss": 0.8105,
+      "step": 12405
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.5800483226484282e-05,
+      "loss": 0.8618,
+      "step": 12410
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.577074651514899e-05,
+      "loss": 0.8118,
+      "step": 12415
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.5741016962138054e-05,
+      "loss": 0.8264,
+      "step": 12420
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.5711294590510055e-05,
+      "loss": 0.8408,
+      "step": 12425
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.5681579423318086e-05,
+      "loss": 0.8159,
+      "step": 12430
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.5651871483609563e-05,
+      "loss": 0.8381,
+      "step": 12435
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.562217079442637e-05,
+      "loss": 0.8443,
+      "step": 12440
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.5592477378804708e-05,
+      "loss": 0.8254,
+      "step": 12445
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.556279125977516e-05,
+      "loss": 0.7998,
+      "step": 12450
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.553311246036267e-05,
+      "loss": 0.8714,
+      "step": 12455
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.5503441003586466e-05,
+      "loss": 0.8355,
+      "step": 12460
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.54737769124601e-05,
+      "loss": 0.8193,
+      "step": 12465
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.5444120209991435e-05,
+      "loss": 0.8381,
+      "step": 12470
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.541447091918256e-05,
+      "loss": 0.8447,
+      "step": 12475
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.5384829063029843e-05,
+      "loss": 0.8568,
+      "step": 12480
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.535519466452388e-05,
+      "loss": 0.869,
+      "step": 12485
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.532556774664947e-05,
+      "loss": 0.8583,
+      "step": 12490
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.5295948332385633e-05,
+      "loss": 0.839,
+      "step": 12495
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.5266336444705563e-05,
+      "loss": 0.8515,
+      "step": 12500
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.5236732106576568e-05,
+      "loss": 0.829,
+      "step": 12505
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.5207135340960204e-05,
+      "loss": 0.8589,
+      "step": 12510
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.517754617081204e-05,
+      "loss": 0.8133,
+      "step": 12515
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.5147964619081848e-05,
+      "loss": 0.8616,
+      "step": 12520
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.511839070871342e-05,
+      "loss": 0.8131,
+      "step": 12525
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.5088824462644654e-05,
+      "loss": 0.8237,
+      "step": 12530
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.5059265903807516e-05,
+      "loss": 0.8603,
+      "step": 12535
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.502971505512798e-05,
+      "loss": 0.8521,
+      "step": 12540
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.5000171939526054e-05,
+      "loss": 0.861,
+      "step": 12545
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.4970636579915754e-05,
+      "loss": 0.8394,
+      "step": 12550
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.4941108999205065e-05,
+      "loss": 0.8635,
+      "step": 12555
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.4911589220295963e-05,
+      "loss": 0.8633,
+      "step": 12560
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.4882077266084342e-05,
+      "loss": 0.784,
+      "step": 12565
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.485257315946005e-05,
+      "loss": 0.8433,
+      "step": 12570
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.4823076923306842e-05,
+      "loss": 0.8786,
+      "step": 12575
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.479358858050237e-05,
+      "loss": 0.8432,
+      "step": 12580
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.476410815391814e-05,
+      "loss": 0.8327,
+      "step": 12585
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.473463566641957e-05,
+      "loss": 0.8006,
+      "step": 12590
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.470517114086586e-05,
+      "loss": 0.8365,
+      "step": 12595
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.4675714600110097e-05,
+      "loss": 0.8394,
+      "step": 12600
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.464626606699912e-05,
+      "loss": 0.8411,
+      "step": 12605
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.461682556437357e-05,
+      "loss": 0.866,
+      "step": 12610
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.45873931150679e-05,
+      "loss": 0.7933,
+      "step": 12615
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.455796874191026e-05,
+      "loss": 0.8575,
+      "step": 12620
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.4528552467722564e-05,
+      "loss": 0.8636,
+      "step": 12625
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.449914431532046e-05,
+      "loss": 0.8061,
+      "step": 12630
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.4469744307513252e-05,
+      "loss": 0.8313,
+      "step": 12635
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.4440352467103975e-05,
+      "loss": 0.8286,
+      "step": 12640
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.4410968816889304e-05,
+      "loss": 0.8829,
+      "step": 12645
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.4381593379659555e-05,
+      "loss": 0.814,
+      "step": 12650
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.4352226178198695e-05,
+      "loss": 0.8134,
+      "step": 12655
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.43228672352843e-05,
+      "loss": 0.8248,
+      "step": 12660
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.4293516573687495e-05,
+      "loss": 0.8408,
+      "step": 12665
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.426417421617307e-05,
+      "loss": 0.8726,
+      "step": 12670
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.423484018549928e-05,
+      "loss": 0.8307,
+      "step": 12675
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.420551450441801e-05,
+      "loss": 0.8432,
+      "step": 12680
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.4176197195674588e-05,
+      "loss": 0.7945,
+      "step": 12685
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.4146888282007895e-05,
+      "loss": 0.8713,
+      "step": 12690
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.4117587786150298e-05,
+      "loss": 0.8465,
+      "step": 12695
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.408829573082763e-05,
+      "loss": 0.8143,
+      "step": 12700
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.4059012138759156e-05,
+      "loss": 0.8416,
+      "step": 12705
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.4029737032657622e-05,
+      "loss": 0.8523,
+      "step": 12710
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.4000470435229143e-05,
+      "loss": 0.8397,
+      "step": 12715
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.3971212369173274e-05,
+      "loss": 0.8281,
+      "step": 12720
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.394196285718294e-05,
+      "loss": 0.8238,
+      "step": 12725
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.39127219219444e-05,
+      "loss": 0.8601,
+      "step": 12730
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.3883489586137322e-05,
+      "loss": 0.8513,
+      "step": 12735
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.385426587243465e-05,
+      "loss": 0.8213,
+      "step": 12740
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.3825050803502664e-05,
+      "loss": 0.8313,
+      "step": 12745
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.379584440200095e-05,
+      "loss": 0.8283,
+      "step": 12750
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.3766646690582333e-05,
+      "loss": 0.8667,
+      "step": 12755
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.3737457691892957e-05,
+      "loss": 0.8112,
+      "step": 12760
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.3708277428572154e-05,
+      "loss": 0.8571,
+      "step": 12765
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.367910592325249e-05,
+      "loss": 0.8583,
+      "step": 12770
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.3649943198559776e-05,
+      "loss": 0.8567,
+      "step": 12775
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.3620789277112967e-05,
+      "loss": 0.8188,
+      "step": 12780
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.3591644181524205e-05,
+      "loss": 0.8456,
+      "step": 12785
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.356250793439881e-05,
+      "loss": 0.8537,
+      "step": 12790
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.35333805583352e-05,
+      "loss": 0.8539,
+      "step": 12795
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.3504262075924944e-05,
+      "loss": 0.8503,
+      "step": 12800
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.3475152509752705e-05,
+      "loss": 0.8593,
+      "step": 12805
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.3446051882396205e-05,
+      "loss": 0.8268,
+      "step": 12810
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.3416960216426283e-05,
+      "loss": 0.8087,
+      "step": 12815
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.338787753440676e-05,
+      "loss": 0.8284,
+      "step": 12820
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.3358803858894575e-05,
+      "loss": 0.8652,
+      "step": 12825
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.3329739212439605e-05,
+      "loss": 0.827,
+      "step": 12830
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.3300683617584745e-05,
+      "loss": 0.8771,
+      "step": 12835
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.3271637096865906e-05,
+      "loss": 0.8609,
+      "step": 12840
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.3242599672811915e-05,
+      "loss": 0.8302,
+      "step": 12845
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.321357136794455e-05,
+      "loss": 0.8234,
+      "step": 12850
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.3184552204778553e-05,
+      "loss": 0.8284,
+      "step": 12855
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.3155542205821532e-05,
+      "loss": 0.9167,
+      "step": 12860
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.3126541393574013e-05,
+      "loss": 0.8782,
+      "step": 12865
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.3097549790529383e-05,
+      "loss": 0.8322,
+      "step": 12870
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.3068567419173893e-05,
+      "loss": 0.8521,
+      "step": 12875
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.303959430198664e-05,
+      "loss": 0.8777,
+      "step": 12880
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.3010630461439533e-05,
+      "loss": 0.8838,
+      "step": 12885
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.298167591999726e-05,
+      "loss": 0.8227,
+      "step": 12890
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.2952730700117376e-05,
+      "loss": 0.8338,
+      "step": 12895
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.2923794824250108e-05,
+      "loss": 0.8423,
+      "step": 12900
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.2894868314838524e-05,
+      "loss": 0.7971,
+      "step": 12905
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.286595119431836e-05,
+      "loss": 0.8753,
+      "step": 12910
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.28370434851181e-05,
+      "loss": 0.8201,
+      "step": 12915
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.2808145209658926e-05,
+      "loss": 0.8294,
+      "step": 12920
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.2779256390354707e-05,
+      "loss": 0.8715,
+      "step": 12925
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.2750377049611962e-05,
+      "loss": 0.8337,
+      "step": 12930
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.2721507209829883e-05,
+      "loss": 0.8291,
+      "step": 12935
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.2692646893400262e-05,
+      "loss": 0.845,
+      "step": 12940
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.2663796122707532e-05,
+      "loss": 0.8107,
+      "step": 12945
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.2634954920128713e-05,
+      "loss": 0.8162,
+      "step": 12950
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.2606123308033387e-05,
+      "loss": 0.8305,
+      "step": 12955
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.2577301308783737e-05,
+      "loss": 0.8495,
+      "step": 12960
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.2548488944734457e-05,
+      "loss": 0.8045,
+      "step": 12965
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.251968623823275e-05,
+      "loss": 0.8253,
+      "step": 12970
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.24908932116184e-05,
+      "loss": 0.8193,
+      "step": 12975
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.246210988722359e-05,
+      "loss": 0.8294,
+      "step": 12980
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.2433336287373077e-05,
+      "loss": 0.8152,
+      "step": 12985
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.240457243438399e-05,
+      "loss": 0.8546,
+      "step": 12990
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.2375818350565937e-05,
+      "loss": 0.8469,
+      "step": 12995
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.2347074058220952e-05,
+      "loss": 0.8354,
+      "step": 13000
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.2318339579643473e-05,
+      "loss": 0.8419,
+      "step": 13005
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.22896149371203e-05,
+      "loss": 0.8514,
+      "step": 13010
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.226090015293065e-05,
+      "loss": 0.8578,
+      "step": 13015
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.2232195249346046e-05,
+      "loss": 0.8191,
+      "step": 13020
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.220350024863039e-05,
+      "loss": 0.8505,
+      "step": 13025
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.217481517303987e-05,
+      "loss": 0.8134,
+      "step": 13030
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.214614004482299e-05,
+      "loss": 0.84,
+      "step": 13035
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.2117474886220542e-05,
+      "loss": 0.8054,
+      "step": 13040
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.208881971946559e-05,
+      "loss": 0.8322,
+      "step": 13045
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.2060174566783408e-05,
+      "loss": 0.8477,
+      "step": 13050
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.2031539450391576e-05,
+      "loss": 0.8236,
+      "step": 13055
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.2002914392499808e-05,
+      "loss": 0.8457,
+      "step": 13060
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.1974299415310094e-05,
+      "loss": 0.8305,
+      "step": 13065
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.1945694541016544e-05,
+      "loss": 0.8446,
+      "step": 13070
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.191709979180546e-05,
+      "loss": 0.8806,
+      "step": 13075
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.1888515189855296e-05,
+      "loss": 0.8299,
+      "step": 13080
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.1859940757336613e-05,
+      "loss": 0.8479,
+      "step": 13085
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.1831376516412105e-05,
+      "loss": 0.8423,
+      "step": 13090
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.1802822489236558e-05,
+      "loss": 0.8337,
+      "step": 13095
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.177427869795682e-05,
+      "loss": 0.8145,
+      "step": 13100
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.1745745164711832e-05,
+      "loss": 0.8518,
+      "step": 13105
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.1717221911632545e-05,
+      "loss": 0.8854,
+      "step": 13110
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.1688708960841947e-05,
+      "loss": 0.8661,
+      "step": 13115
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.1660206334455048e-05,
+      "loss": 0.818,
+      "step": 13120
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.1631714054578852e-05,
+      "loss": 0.855,
+      "step": 13125
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.1603232143312287e-05,
+      "loss": 0.811,
+      "step": 13130
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.157476062274633e-05,
+      "loss": 0.8065,
+      "step": 13135
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.1546299514963805e-05,
+      "loss": 0.8416,
+      "step": 13140
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.1517848842039543e-05,
+      "loss": 0.8071,
+      "step": 13145
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.148940862604021e-05,
+      "loss": 0.8978,
+      "step": 13150
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.14609788890244e-05,
+      "loss": 0.8487,
+      "step": 13155
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.143255965304258e-05,
+      "loss": 0.8789,
+      "step": 13160
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.1404150940137068e-05,
+      "loss": 0.85,
+      "step": 13165
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.1375752772341994e-05,
+      "loss": 0.8303,
+      "step": 13170
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.134736517168336e-05,
+      "loss": 0.8177,
+      "step": 13175
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.1318988160178922e-05,
+      "loss": 0.8002,
+      "step": 13180
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.1290621759838264e-05,
+      "loss": 0.8385,
+      "step": 13185
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.1262265992662708e-05,
+      "loss": 0.8421,
+      "step": 13190
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.1233920880645334e-05,
+      "loss": 0.8395,
+      "step": 13195
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.1205586445770988e-05,
+      "loss": 0.8457,
+      "step": 13200
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.1177262710016204e-05,
+      "loss": 0.8102,
+      "step": 13205
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.1148949695349198e-05,
+      "loss": 0.876,
+      "step": 13210
+    },
+    {
+      "epoch": 2.0,
+      "eval_loss": 0.9647670388221741,
+      "eval_runtime": 235.4788,
+      "eval_samples_per_second": 47.253,
+      "eval_steps_per_second": 5.907,
+      "step": 13212
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.112064742372994e-05,
+      "loss": 0.7942,
+      "step": 13215
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.109235591710999e-05,
+      "loss": 0.7675,
+      "step": 13220
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.1064075197432628e-05,
+      "loss": 0.7613,
+      "step": 13225
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.103580528663271e-05,
+      "loss": 0.8033,
+      "step": 13230
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.100754620663673e-05,
+      "loss": 0.7901,
+      "step": 13235
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.09792979793628e-05,
+      "loss": 0.7968,
+      "step": 13240
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.0951060626720593e-05,
+      "loss": 0.7802,
+      "step": 13245
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.0922834170611337e-05,
+      "loss": 0.7783,
+      "step": 13250
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.0894618632927844e-05,
+      "loss": 0.7619,
+      "step": 13255
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.0866414035554424e-05,
+      "loss": 0.7483,
+      "step": 13260
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.083822040036693e-05,
+      "loss": 0.7439,
+      "step": 13265
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.0810037749232697e-05,
+      "loss": 0.7707,
+      "step": 13270
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.078186610401051e-05,
+      "loss": 0.8015,
+      "step": 13275
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.0753705486550685e-05,
+      "loss": 0.8029,
+      "step": 13280
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.0725555918694937e-05,
+      "loss": 0.7627,
+      "step": 13285
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.0697417422276423e-05,
+      "loss": 0.7631,
+      "step": 13290
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.0669290019119717e-05,
+      "loss": 0.784,
+      "step": 13295
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.064117373104076e-05,
+      "loss": 0.8114,
+      "step": 13300
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.061306857984694e-05,
+      "loss": 0.7681,
+      "step": 13305
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.0584974587336934e-05,
+      "loss": 0.7635,
+      "step": 13310
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.0556891775300788e-05,
+      "loss": 0.7537,
+      "step": 13315
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.05288201655199e-05,
+      "loss": 0.7771,
+      "step": 13320
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.050075977976695e-05,
+      "loss": 0.7655,
+      "step": 13325
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.0472710639805932e-05,
+      "loss": 0.8014,
+      "step": 13330
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.0444672767392105e-05,
+      "loss": 0.7908,
+      "step": 13335
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.0416646184271983e-05,
+      "loss": 0.7747,
+      "step": 13340
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.0388630912183366e-05,
+      "loss": 0.8081,
+      "step": 13345
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.0360626972855227e-05,
+      "loss": 0.7697,
+      "step": 13350
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.0332634388007757e-05,
+      "loss": 0.8012,
+      "step": 13355
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.0304653179352376e-05,
+      "loss": 0.7966,
+      "step": 13360
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.027668336859165e-05,
+      "loss": 0.7623,
+      "step": 13365
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.024872497741933e-05,
+      "loss": 0.7772,
+      "step": 13370
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.0220778027520268e-05,
+      "loss": 0.7635,
+      "step": 13375
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.019284254057048e-05,
+      "loss": 0.7585,
+      "step": 13380
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.0164918538237095e-05,
+      "loss": 0.7507,
+      "step": 13385
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.01370060421783e-05,
+      "loss": 0.7852,
+      "step": 13390
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.0109105074043352e-05,
+      "loss": 0.8033,
+      "step": 13395
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.0081215655472643e-05,
+      "loss": 0.7273,
+      "step": 13400
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.0053337808097517e-05,
+      "loss": 0.7594,
+      "step": 13405
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.0025471553540406e-05,
+      "loss": 0.7589,
+      "step": 13410
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.99976169134147e-05,
+      "loss": 0.7655,
+      "step": 13415
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.996977390932481e-05,
+      "loss": 0.7779,
+      "step": 13420
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.994194256286615e-05,
+      "loss": 0.814,
+      "step": 13425
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9914122895625015e-05,
+      "loss": 0.8077,
+      "step": 13430
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9886314929178705e-05,
+      "loss": 0.7385,
+      "step": 13435
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9858518685095446e-05,
+      "loss": 0.8087,
+      "step": 13440
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9830734184934326e-05,
+      "loss": 0.7804,
+      "step": 13445
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9802961450245357e-05,
+      "loss": 0.7607,
+      "step": 13450
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.977520050256943e-05,
+      "loss": 0.7684,
+      "step": 13455
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9747451363438266e-05,
+      "loss": 0.8272,
+      "step": 13460
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9719714054374452e-05,
+      "loss": 0.7799,
+      "step": 13465
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9691988596891403e-05,
+      "loss": 0.7618,
+      "step": 13470
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.966427501249331e-05,
+      "loss": 0.7822,
+      "step": 13475
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.96365733226752e-05,
+      "loss": 0.783,
+      "step": 13480
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9608883548922806e-05,
+      "loss": 0.7553,
+      "step": 13485
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.958120571271272e-05,
+      "loss": 0.7495,
+      "step": 13490
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9553539835512186e-05,
+      "loss": 0.7961,
+      "step": 13495
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9525885938779197e-05,
+      "loss": 0.7908,
+      "step": 13500
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.949824404396247e-05,
+      "loss": 0.7831,
+      "step": 13505
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.947061417250142e-05,
+      "loss": 0.7846,
+      "step": 13510
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9442996345826092e-05,
+      "loss": 0.7821,
+      "step": 13515
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9415390585357232e-05,
+      "loss": 0.7651,
+      "step": 13520
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9387796912506212e-05,
+      "loss": 0.8145,
+      "step": 13525
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9360215348675044e-05,
+      "loss": 0.8203,
+      "step": 13530
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9332645915256306e-05,
+      "loss": 0.8378,
+      "step": 13535
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9305088633633208e-05,
+      "loss": 0.762,
+      "step": 13540
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9277543525179537e-05,
+      "loss": 0.7911,
+      "step": 13545
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.925001061125961e-05,
+      "loss": 0.8037,
+      "step": 13550
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9222489913228268e-05,
+      "loss": 0.7431,
+      "step": 13555
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.919498145243097e-05,
+      "loss": 0.7587,
+      "step": 13560
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9167485250203576e-05,
+      "loss": 0.772,
+      "step": 13565
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.914000132787251e-05,
+      "loss": 0.788,
+      "step": 13570
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.911252970675462e-05,
+      "loss": 0.7817,
+      "step": 13575
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9085070408157258e-05,
+      "loss": 0.7751,
+      "step": 13580
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9057623453378205e-05,
+      "loss": 0.7628,
+      "step": 13585
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9030188863705635e-05,
+      "loss": 0.7633,
+      "step": 13590
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9002766660418183e-05,
+      "loss": 0.7299,
+      "step": 13595
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.8975356864784855e-05,
+      "loss": 0.7561,
+      "step": 13600
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.8947959498065002e-05,
+      "loss": 0.7588,
+      "step": 13605
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.892057458150839e-05,
+      "loss": 0.7553,
+      "step": 13610
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.8893202136355106e-05,
+      "loss": 0.7945,
+      "step": 13615
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.8865842183835538e-05,
+      "loss": 0.7976,
+      "step": 13620
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.883849474517041e-05,
+      "loss": 0.7745,
+      "step": 13625
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.8811159841570772e-05,
+      "loss": 0.7698,
+      "step": 13630
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.878383749423787e-05,
+      "loss": 0.7868,
+      "step": 13635
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.8756527724363298e-05,
+      "loss": 0.8188,
+      "step": 13640
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.8729230553128804e-05,
+      "loss": 0.7555,
+      "step": 13645
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.8701946001706477e-05,
+      "loss": 0.7703,
+      "step": 13650
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.8674674091258528e-05,
+      "loss": 0.792,
+      "step": 13655
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.8647414842937372e-05,
+      "loss": 0.8058,
+      "step": 13660
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.8620168277885642e-05,
+      "loss": 0.7479,
+      "step": 13665
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.8592934417236127e-05,
+      "loss": 0.7701,
+      "step": 13670
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.8565713282111726e-05,
+      "loss": 0.7867,
+      "step": 13675
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.8538504893625492e-05,
+      "loss": 0.7469,
+      "step": 13680
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.8511309272880607e-05,
+      "loss": 0.7979,
+      "step": 13685
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.8484126440970342e-05,
+      "loss": 0.7651,
+      "step": 13690
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.845695641897801e-05,
+      "loss": 0.7818,
+      "step": 13695
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.8429799227977036e-05,
+      "loss": 0.7601,
+      "step": 13700
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.8402654889030897e-05,
+      "loss": 0.7522,
+      "step": 13705
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.837552342319306e-05,
+      "loss": 0.7547,
+      "step": 13710
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.8348404851507e-05,
+      "loss": 0.7369,
+      "step": 13715
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.8321299195006277e-05,
+      "loss": 0.779,
+      "step": 13720
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.829420647471434e-05,
+      "loss": 0.7483,
+      "step": 13725
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.8267126711644666e-05,
+      "loss": 0.7361,
+      "step": 13730
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.8240059926800635e-05,
+      "loss": 0.8345,
+      "step": 13735
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.8213006141175584e-05,
+      "loss": 0.7489,
+      "step": 13740
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.8185965375752793e-05,
+      "loss": 0.782,
+      "step": 13745
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.815893765150539e-05,
+      "loss": 0.8044,
+      "step": 13750
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.8131922989396426e-05,
+      "loss": 0.772,
+      "step": 13755
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.810492141037882e-05,
+      "loss": 0.7937,
+      "step": 13760
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.807793293539532e-05,
+      "loss": 0.7789,
+      "step": 13765
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.805095758537852e-05,
+      "loss": 0.7773,
+      "step": 13770
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.8023995381250866e-05,
+      "loss": 0.7953,
+      "step": 13775
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.799704634392455e-05,
+      "loss": 0.7629,
+      "step": 13780
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.797011049430159e-05,
+      "loss": 0.8045,
+      "step": 13785
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.794318785327378e-05,
+      "loss": 0.772,
+      "step": 13790
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.791627844172266e-05,
+      "loss": 0.7337,
+      "step": 13795
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.788938228051949e-05,
+      "loss": 0.7725,
+      "step": 13800
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.786249939052525e-05,
+      "loss": 0.8028,
+      "step": 13805
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.7835629792590696e-05,
+      "loss": 0.7716,
+      "step": 13810
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.78087735075562e-05,
+      "loss": 0.7849,
+      "step": 13815
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.7781930556251814e-05,
+      "loss": 0.819,
+      "step": 13820
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.775510095949728e-05,
+      "loss": 0.7916,
+      "step": 13825
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.772828473810198e-05,
+      "loss": 0.7827,
+      "step": 13830
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.7701481912864908e-05,
+      "loss": 0.7534,
+      "step": 13835
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.767469250457465e-05,
+      "loss": 0.7991,
+      "step": 13840
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.764791653400942e-05,
+      "loss": 0.7663,
+      "step": 13845
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.762115402193701e-05,
+      "loss": 0.7499,
+      "step": 13850
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.7594404989114738e-05,
+      "loss": 0.8095,
+      "step": 13855
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.7567669456289472e-05,
+      "loss": 0.7653,
+      "step": 13860
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.7540947444197673e-05,
+      "loss": 0.746,
+      "step": 13865
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.7514238973565232e-05,
+      "loss": 0.7796,
+      "step": 13870
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.748754406510758e-05,
+      "loss": 0.7271,
+      "step": 13875
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.7460862739529633e-05,
+      "loss": 0.7713,
+      "step": 13880
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.743419501752574e-05,
+      "loss": 0.7476,
+      "step": 13885
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.740754091977975e-05,
+      "loss": 0.7971,
+      "step": 13890
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.7380900466964878e-05,
+      "loss": 0.8161,
+      "step": 13895
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.7354273679743812e-05,
+      "loss": 0.792,
+      "step": 13900
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.7327660578768628e-05,
+      "loss": 0.7854,
+      "step": 13905
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.7301061184680758e-05,
+      "loss": 0.7864,
+      "step": 13910
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.727447551811104e-05,
+      "loss": 0.7807,
+      "step": 13915
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.7247903599679655e-05,
+      "loss": 0.7855,
+      "step": 13920
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.72213454499961e-05,
+      "loss": 0.797,
+      "step": 13925
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.7194801089659206e-05,
+      "loss": 0.7653,
+      "step": 13930
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.7168270539257133e-05,
+      "loss": 0.7638,
+      "step": 13935
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.7141753819367283e-05,
+      "loss": 0.7442,
+      "step": 13940
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.7115250950556377e-05,
+      "loss": 0.7532,
+      "step": 13945
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.7088761953380334e-05,
+      "loss": 0.7777,
+      "step": 13950
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.70622868483844e-05,
+      "loss": 0.7594,
+      "step": 13955
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.703582565610298e-05,
+      "loss": 0.7646,
+      "step": 13960
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.700937839705967e-05,
+      "loss": 0.7558,
+      "step": 13965
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.6982945091767352e-05,
+      "loss": 0.7792,
+      "step": 13970
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.6956525760728002e-05,
+      "loss": 0.7512,
+      "step": 13975
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.6930120424432764e-05,
+      "loss": 0.8168,
+      "step": 13980
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.690372910336197e-05,
+      "loss": 0.7462,
+      "step": 13985
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.6877351817985048e-05,
+      "loss": 0.7394,
+      "step": 13990
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.685098858876057e-05,
+      "loss": 0.7813,
+      "step": 13995
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.6824639436136158e-05,
+      "loss": 0.7906,
+      "step": 14000
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.6798304380548555e-05,
+      "loss": 0.741,
+      "step": 14005
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.677198344242358e-05,
+      "loss": 0.806,
+      "step": 14010
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.674567664217606e-05,
+      "loss": 0.8082,
+      "step": 14015
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.6719384000209855e-05,
+      "loss": 0.805,
+      "step": 14020
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.669310553691792e-05,
+      "loss": 0.7743,
+      "step": 14025
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.6666841272682118e-05,
+      "loss": 0.812,
+      "step": 14030
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.664059122787337e-05,
+      "loss": 0.7716,
+      "step": 14035
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.661435542285151e-05,
+      "loss": 0.761,
+      "step": 14040
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.6588133877965372e-05,
+      "loss": 0.7699,
+      "step": 14045
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.6561926613552726e-05,
+      "loss": 0.7831,
+      "step": 14050
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.6535733649940224e-05,
+      "loss": 0.7636,
+      "step": 14055
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.650955500744347e-05,
+      "loss": 0.7647,
+      "step": 14060
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.6483390706366966e-05,
+      "loss": 0.7799,
+      "step": 14065
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.6457240767004036e-05,
+      "loss": 0.7792,
+      "step": 14070
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.643110520963692e-05,
+      "loss": 0.7487,
+      "step": 14075
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.6404984054536693e-05,
+      "loss": 0.7592,
+      "step": 14080
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.6378877321963224e-05,
+      "loss": 0.7953,
+      "step": 14085
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.635278503216523e-05,
+      "loss": 0.7493,
+      "step": 14090
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.6326707205380234e-05,
+      "loss": 0.7741,
+      "step": 14095
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.6300643861834498e-05,
+      "loss": 0.7975,
+      "step": 14100
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.6274595021743103e-05,
+      "loss": 0.7823,
+      "step": 14105
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.6248560705309817e-05,
+      "loss": 0.7815,
+      "step": 14110
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.6222540932727232e-05,
+      "loss": 0.7921,
+      "step": 14115
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.619653572417658e-05,
+      "loss": 0.7622,
+      "step": 14120
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.617054509982782e-05,
+      "loss": 0.7863,
+      "step": 14125
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.614456907983961e-05,
+      "loss": 0.7492,
+      "step": 14130
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.611860768435929e-05,
+      "loss": 0.7658,
+      "step": 14135
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.6092660933522824e-05,
+      "loss": 0.7826,
+      "step": 14140
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.6066728847454844e-05,
+      "loss": 0.7806,
+      "step": 14145
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.60408114462686e-05,
+      "loss": 0.7681,
+      "step": 14150
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.6014908750065964e-05,
+      "loss": 0.7825,
+      "step": 14155
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.5989020778937367e-05,
+      "loss": 0.7776,
+      "step": 14160
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.5963147552961856e-05,
+      "loss": 0.7871,
+      "step": 14165
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.593728909220704e-05,
+      "loss": 0.809,
+      "step": 14170
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.5911445416729058e-05,
+      "loss": 0.7611,
+      "step": 14175
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.588561654657255e-05,
+      "loss": 0.7625,
+      "step": 14180
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.585980250177077e-05,
+      "loss": 0.7983,
+      "step": 14185
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.583400330234537e-05,
+      "loss": 0.7669,
+      "step": 14190
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.5808218968306562e-05,
+      "loss": 0.8058,
+      "step": 14195
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.5782449519652976e-05,
+      "loss": 0.7793,
+      "step": 14200
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.5756694976371725e-05,
+      "loss": 0.7881,
+      "step": 14205
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.5730955358438362e-05,
+      "loss": 0.7488,
+      "step": 14210
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.5705230685816843e-05,
+      "loss": 0.8123,
+      "step": 14215
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.567952097845955e-05,
+      "loss": 0.787,
+      "step": 14220
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.565382625630727e-05,
+      "loss": 0.7834,
+      "step": 14225
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.5628146539289123e-05,
+      "loss": 0.7753,
+      "step": 14230
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.5602481847322636e-05,
+      "loss": 0.7827,
+      "step": 14235
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.557683220031367e-05,
+      "loss": 0.7807,
+      "step": 14240
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.5551197618156388e-05,
+      "loss": 0.7918,
+      "step": 14245
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.552557812073331e-05,
+      "loss": 0.742,
+      "step": 14250
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.549997372791524e-05,
+      "loss": 0.7558,
+      "step": 14255
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.5474384459561264e-05,
+      "loss": 0.7861,
+      "step": 14260
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.544881033551874e-05,
+      "loss": 0.764,
+      "step": 14265
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.542325137562325e-05,
+      "loss": 0.7693,
+      "step": 14270
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.539770759969869e-05,
+      "loss": 0.7784,
+      "step": 14275
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.5372179027557103e-05,
+      "loss": 0.7831,
+      "step": 14280
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.5346665678998762e-05,
+      "loss": 0.7763,
+      "step": 14285
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.5321167573812155e-05,
+      "loss": 0.77,
+      "step": 14290
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.529568473177392e-05,
+      "loss": 0.7558,
+      "step": 14295
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.5270217172648887e-05,
+      "loss": 0.7846,
+      "step": 14300
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.5244764916189996e-05,
+      "loss": 0.7601,
+      "step": 14305
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.5219327982138332e-05,
+      "loss": 0.7787,
+      "step": 14310
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.5193906390223124e-05,
+      "loss": 0.8051,
+      "step": 14315
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.5168500160161642e-05,
+      "loss": 0.7905,
+      "step": 14320
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.5143109311659292e-05,
+      "loss": 0.8085,
+      "step": 14325
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.511773386440955e-05,
+      "loss": 0.7412,
+      "step": 14330
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.5092373838093901e-05,
+      "loss": 0.7862,
+      "step": 14335
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.5067029252381909e-05,
+      "loss": 0.7933,
+      "step": 14340
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.5041700126931169e-05,
+      "loss": 0.7656,
+      "step": 14345
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.5016386481387236e-05,
+      "loss": 0.7879,
+      "step": 14350
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.499108833538372e-05,
+      "loss": 0.7909,
+      "step": 14355
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.4965805708542153e-05,
+      "loss": 0.7943,
+      "step": 14360
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.4940538620472069e-05,
+      "loss": 0.7784,
+      "step": 14365
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.4915287090770948e-05,
+      "loss": 0.7807,
+      "step": 14370
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.4890051139024169e-05,
+      "loss": 0.797,
+      "step": 14375
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.4864830784805064e-05,
+      "loss": 0.7645,
+      "step": 14380
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.4839626047674867e-05,
+      "loss": 0.7839,
+      "step": 14385
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.4814436947182659e-05,
+      "loss": 0.7661,
+      "step": 14390
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.4789263502865439e-05,
+      "loss": 0.8029,
+      "step": 14395
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.4764105734248056e-05,
+      "loss": 0.7559,
+      "step": 14400
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.473896366084316e-05,
+      "loss": 0.7792,
+      "step": 14405
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.471383730215128e-05,
+      "loss": 0.7481,
+      "step": 14410
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.4688726677660722e-05,
+      "loss": 0.8084,
+      "step": 14415
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.4663631806847622e-05,
+      "loss": 0.7659,
+      "step": 14420
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.4638552709175858e-05,
+      "loss": 0.7424,
+      "step": 14425
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.4613489404097064e-05,
+      "loss": 0.8281,
+      "step": 14430
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.4588441911050711e-05,
+      "loss": 0.7788,
+      "step": 14435
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.4563410249463916e-05,
+      "loss": 0.7486,
+      "step": 14440
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.453839443875154e-05,
+      "loss": 0.7913,
+      "step": 14445
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.4513394498316167e-05,
+      "loss": 0.7776,
+      "step": 14450
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.4488410447548068e-05,
+      "loss": 0.7617,
+      "step": 14455
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.446344230582519e-05,
+      "loss": 0.8245,
+      "step": 14460
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.443849009251312e-05,
+      "loss": 0.776,
+      "step": 14465
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.4413553826965114e-05,
+      "loss": 0.739,
+      "step": 14470
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.4388633528522064e-05,
+      "loss": 0.7727,
+      "step": 14475
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.4363729216512455e-05,
+      "loss": 0.7764,
+      "step": 14480
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.4338840910252354e-05,
+      "loss": 0.7683,
+      "step": 14485
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.4313968629045494e-05,
+      "loss": 0.8012,
+      "step": 14490
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.4289112392183086e-05,
+      "loss": 0.7618,
+      "step": 14495
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.426427221894396e-05,
+      "loss": 0.7956,
+      "step": 14500
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.4239448128594468e-05,
+      "loss": 0.772,
+      "step": 14505
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.4214640140388463e-05,
+      "loss": 0.7919,
+      "step": 14510
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.4189848273567356e-05,
+      "loss": 0.7584,
+      "step": 14515
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.4165072547360001e-05,
+      "loss": 0.7827,
+      "step": 14520
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.4140312980982782e-05,
+      "loss": 0.7705,
+      "step": 14525
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.411556959363953e-05,
+      "loss": 0.7254,
+      "step": 14530
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.4090842404521505e-05,
+      "loss": 0.7673,
+      "step": 14535
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.406613143280744e-05,
+      "loss": 0.7962,
+      "step": 14540
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.404143669766348e-05,
+      "loss": 0.7638,
+      "step": 14545
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.401675821824315e-05,
+      "loss": 0.7829,
+      "step": 14550
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3992096013687392e-05,
+      "loss": 0.7649,
+      "step": 14555
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3967450103124541e-05,
+      "loss": 0.7601,
+      "step": 14560
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3942820505670244e-05,
+      "loss": 0.711,
+      "step": 14565
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3918207240427553e-05,
+      "loss": 0.7641,
+      "step": 14570
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3893610326486782e-05,
+      "loss": 0.7598,
+      "step": 14575
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3869029782925658e-05,
+      "loss": 0.7879,
+      "step": 14580
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.384446562880913e-05,
+      "loss": 0.7794,
+      "step": 14585
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3819917883189458e-05,
+      "loss": 0.8133,
+      "step": 14590
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3795386565106184e-05,
+      "loss": 0.7968,
+      "step": 14595
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3770871693586118e-05,
+      "loss": 0.7709,
+      "step": 14600
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3746373287643282e-05,
+      "loss": 0.8152,
+      "step": 14605
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3721891366278954e-05,
+      "loss": 0.7642,
+      "step": 14610
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.369742594848162e-05,
+      "loss": 0.7801,
+      "step": 14615
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3672977053226973e-05,
+      "loss": 0.7725,
+      "step": 14620
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.364854469947786e-05,
+      "loss": 0.728,
+      "step": 14625
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.362412890618433e-05,
+      "loss": 0.766,
+      "step": 14630
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3599729692283594e-05,
+      "loss": 0.8108,
+      "step": 14635
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.357534707669997e-05,
+      "loss": 0.8021,
+      "step": 14640
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3550981078344897e-05,
+      "loss": 0.8057,
+      "step": 14645
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3526631716116997e-05,
+      "loss": 0.7581,
+      "step": 14650
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3502299008901903e-05,
+      "loss": 0.7765,
+      "step": 14655
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3477982975572389e-05,
+      "loss": 0.723,
+      "step": 14660
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3453683634988257e-05,
+      "loss": 0.7885,
+      "step": 14665
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3429401005996383e-05,
+      "loss": 0.8053,
+      "step": 14670
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.340513510743069e-05,
+      "loss": 0.7706,
+      "step": 14675
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3380885958112094e-05,
+      "loss": 0.8157,
+      "step": 14680
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.335665357684855e-05,
+      "loss": 0.801,
+      "step": 14685
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3332437982435003e-05,
+      "loss": 0.7523,
+      "step": 14690
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3308239193653349e-05,
+      "loss": 0.7572,
+      "step": 14695
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.328405722927248e-05,
+      "loss": 0.8207,
+      "step": 14700
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.3259892108048251e-05,
+      "loss": 0.8142,
+      "step": 14705
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.3235743848723397e-05,
+      "loss": 0.7761,
+      "step": 14710
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.3211612470027627e-05,
+      "loss": 0.7693,
+      "step": 14715
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.3187497990677538e-05,
+      "loss": 0.7408,
+      "step": 14720
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.3163400429376637e-05,
+      "loss": 0.7652,
+      "step": 14725
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.3139319804815279e-05,
+      "loss": 0.8159,
+      "step": 14730
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.3115256135670676e-05,
+      "loss": 0.8051,
+      "step": 14735
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.3091209440606955e-05,
+      "loss": 0.7896,
+      "step": 14740
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.3067179738275011e-05,
+      "loss": 0.8008,
+      "step": 14745
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.3043167047312567e-05,
+      "loss": 0.7962,
+      "step": 14750
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.3019171386344185e-05,
+      "loss": 0.7714,
+      "step": 14755
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.299519277398119e-05,
+      "loss": 0.8067,
+      "step": 14760
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2971231228821716e-05,
+      "loss": 0.7892,
+      "step": 14765
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2947286769450603e-05,
+      "loss": 0.8176,
+      "step": 14770
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2923359414439493e-05,
+      "loss": 0.7986,
+      "step": 14775
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2899449182346747e-05,
+      "loss": 0.7663,
+      "step": 14780
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2875556091717418e-05,
+      "loss": 0.7808,
+      "step": 14785
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2851680161083301e-05,
+      "loss": 0.7559,
+      "step": 14790
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2827821408962874e-05,
+      "loss": 0.8047,
+      "step": 14795
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.280397985386126e-05,
+      "loss": 0.7734,
+      "step": 14800
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.278015551427028e-05,
+      "loss": 0.7652,
+      "step": 14805
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2756348408668408e-05,
+      "loss": 0.8046,
+      "step": 14810
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2732558555520696e-05,
+      "loss": 0.7699,
+      "step": 14815
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2708785973278885e-05,
+      "loss": 0.7692,
+      "step": 14820
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2685030680381256e-05,
+      "loss": 0.7844,
+      "step": 14825
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.266129269525273e-05,
+      "loss": 0.7811,
+      "step": 14830
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2637572036304795e-05,
+      "loss": 0.7671,
+      "step": 14835
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.261386872193546e-05,
+      "loss": 0.8119,
+      "step": 14840
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2590182770529332e-05,
+      "loss": 0.7814,
+      "step": 14845
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.256651420045754e-05,
+      "loss": 0.794,
+      "step": 14850
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2542863030077698e-05,
+      "loss": 0.8047,
+      "step": 14855
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2519229277733963e-05,
+      "loss": 0.7706,
+      "step": 14860
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.249561296175698e-05,
+      "loss": 0.7855,
+      "step": 14865
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2472014100463835e-05,
+      "loss": 0.7607,
+      "step": 14870
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2448432712158107e-05,
+      "loss": 0.7572,
+      "step": 14875
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2424868815129825e-05,
+      "loss": 0.7909,
+      "step": 14880
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2401322427655449e-05,
+      "loss": 0.7828,
+      "step": 14885
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2377793567997837e-05,
+      "loss": 0.7427,
+      "step": 14890
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2354282254406246e-05,
+      "loss": 0.7891,
+      "step": 14895
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2330788505116393e-05,
+      "loss": 0.7779,
+      "step": 14900
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2307312338350292e-05,
+      "loss": 0.8025,
+      "step": 14905
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2283853772316353e-05,
+      "loss": 0.7683,
+      "step": 14910
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2260412825209336e-05,
+      "loss": 0.7927,
+      "step": 14915
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2236989515210333e-05,
+      "loss": 0.7753,
+      "step": 14920
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2213583860486766e-05,
+      "loss": 0.7493,
+      "step": 14925
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2190195879192339e-05,
+      "loss": 0.7642,
+      "step": 14930
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2166825589467064e-05,
+      "loss": 0.7687,
+      "step": 14935
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2143473009437246e-05,
+      "loss": 0.752,
+      "step": 14940
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2120138157215412e-05,
+      "loss": 0.8122,
+      "step": 14945
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2096821050900383e-05,
+      "loss": 0.7332,
+      "step": 14950
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2073521708577202e-05,
+      "loss": 0.7562,
+      "step": 14955
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.205024014831711e-05,
+      "loss": 0.7125,
+      "step": 14960
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.202697638817759e-05,
+      "loss": 0.7733,
+      "step": 14965
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2003730446202313e-05,
+      "loss": 0.7285,
+      "step": 14970
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1980502340421105e-05,
+      "loss": 0.7805,
+      "step": 14975
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1957292088849993e-05,
+      "loss": 0.8042,
+      "step": 14980
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1934099709491114e-05,
+      "loss": 0.7689,
+      "step": 14985
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.191092522033278e-05,
+      "loss": 0.7604,
+      "step": 14990
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1887768639349428e-05,
+      "loss": 0.7598,
+      "step": 14995
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1864629984501566e-05,
+      "loss": 0.7855,
+      "step": 15000
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.184150927373584e-05,
+      "loss": 0.7883,
+      "step": 15005
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.181840652498497e-05,
+      "loss": 0.7488,
+      "step": 15010
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1795321756167714e-05,
+      "loss": 0.7602,
+      "step": 15015
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1772254985188917e-05,
+      "loss": 0.7778,
+      "step": 15020
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1749206229939471e-05,
+      "loss": 0.794,
+      "step": 15025
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1726175508296255e-05,
+      "loss": 0.7515,
+      "step": 15030
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.170316283812219e-05,
+      "loss": 0.7571,
+      "step": 15035
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1680168237266195e-05,
+      "loss": 0.7385,
+      "step": 15040
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1657191723563182e-05,
+      "loss": 0.7744,
+      "step": 15045
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1634233314834011e-05,
+      "loss": 0.7723,
+      "step": 15050
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1611293028885486e-05,
+      "loss": 0.7647,
+      "step": 15055
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1588370883510425e-05,
+      "loss": 0.7434,
+      "step": 15060
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1565466896487508e-05,
+      "loss": 0.7755,
+      "step": 15065
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1542581085581343e-05,
+      "loss": 0.8091,
+      "step": 15070
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1519713468542461e-05,
+      "loss": 0.7777,
+      "step": 15075
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1496864063107273e-05,
+      "loss": 0.755,
+      "step": 15080
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1474032886998076e-05,
+      "loss": 0.7686,
+      "step": 15085
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1451219957922987e-05,
+      "loss": 0.7793,
+      "step": 15090
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1428425293576018e-05,
+      "loss": 0.7844,
+      "step": 15095
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1405648911637003e-05,
+      "loss": 0.7678,
+      "step": 15100
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1382890829771578e-05,
+      "loss": 0.7607,
+      "step": 15105
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1360151065631172e-05,
+      "loss": 0.7389,
+      "step": 15110
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1337429636853072e-05,
+      "loss": 0.7648,
+      "step": 15115
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1314726561060277e-05,
+      "loss": 0.7869,
+      "step": 15120
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1292041855861592e-05,
+      "loss": 0.7853,
+      "step": 15125
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1269375538851538e-05,
+      "loss": 0.7407,
+      "step": 15130
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1246727627610405e-05,
+      "loss": 0.7474,
+      "step": 15135
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1224098139704206e-05,
+      "loss": 0.7507,
+      "step": 15140
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1201487092684636e-05,
+      "loss": 0.7663,
+      "step": 15145
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1178894504089113e-05,
+      "loss": 0.7906,
+      "step": 15150
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.115632039144074e-05,
+      "loss": 0.7582,
+      "step": 15155
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.113376477224826e-05,
+      "loss": 0.752,
+      "step": 15160
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1111227664006112e-05,
+      "loss": 0.7658,
+      "step": 15165
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1088709084194359e-05,
+      "loss": 0.7895,
+      "step": 15170
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1066209050278679e-05,
+      "loss": 0.7721,
+      "step": 15175
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1043727579710387e-05,
+      "loss": 0.7748,
+      "step": 15180
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1021264689926391e-05,
+      "loss": 0.7784,
+      "step": 15185
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.0998820398349206e-05,
+      "loss": 0.7414,
+      "step": 15190
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.0976394722386885e-05,
+      "loss": 0.757,
+      "step": 15195
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.0953987679433047e-05,
+      "loss": 0.8107,
+      "step": 15200
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.0931599286866916e-05,
+      "loss": 0.7771,
+      "step": 15205
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.0909229562053193e-05,
+      "loss": 0.8025,
+      "step": 15210
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.0886878522342095e-05,
+      "loss": 0.7794,
+      "step": 15215
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.0864546185069386e-05,
+      "loss": 0.806,
+      "step": 15220
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.0842232567556302e-05,
+      "loss": 0.78,
+      "step": 15225
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.0819937687109582e-05,
+      "loss": 0.7348,
+      "step": 15230
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.0797661561021381e-05,
+      "loss": 0.7963,
+      "step": 15235
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.0775404206569358e-05,
+      "loss": 0.7529,
+      "step": 15240
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.0753165641016606e-05,
+      "loss": 0.7762,
+      "step": 15245
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.0730945881611611e-05,
+      "loss": 0.7599,
+      "step": 15250
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.0708744945588312e-05,
+      "loss": 0.7709,
+      "step": 15255
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.0686562850166037e-05,
+      "loss": 0.7787,
+      "step": 15260
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.0664399612549479e-05,
+      "loss": 0.7785,
+      "step": 15265
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.0642255249928728e-05,
+      "loss": 0.7469,
+      "step": 15270
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.0620129779479242e-05,
+      "loss": 0.7626,
+      "step": 15275
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.0598023218361794e-05,
+      "loss": 0.7667,
+      "step": 15280
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.0575935583722524e-05,
+      "loss": 0.7883,
+      "step": 15285
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.0553866892692862e-05,
+      "loss": 0.8096,
+      "step": 15290
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.053181716238957e-05,
+      "loss": 0.804,
+      "step": 15295
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.0509786409914704e-05,
+      "loss": 0.7837,
+      "step": 15300
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.0487774652355567e-05,
+      "loss": 0.7851,
+      "step": 15305
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.0465781906784768e-05,
+      "loss": 0.7663,
+      "step": 15310
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.0443808190260166e-05,
+      "loss": 0.7544,
+      "step": 15315
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.0421853519824827e-05,
+      "loss": 0.7498,
+      "step": 15320
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.0399917912507076e-05,
+      "loss": 0.808,
+      "step": 15325
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.0378001385320464e-05,
+      "loss": 0.7491,
+      "step": 15330
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.0356103955263692e-05,
+      "loss": 0.776,
+      "step": 15335
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.0334225639320694e-05,
+      "loss": 0.77,
+      "step": 15340
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.0312366454460564e-05,
+      "loss": 0.7437,
+      "step": 15345
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.0290526417637577e-05,
+      "loss": 0.8027,
+      "step": 15350
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.0268705545791122e-05,
+      "loss": 0.7455,
+      "step": 15355
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.0246903855845715e-05,
+      "loss": 0.7604,
+      "step": 15360
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.0225121364711071e-05,
+      "loss": 0.7585,
+      "step": 15365
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.0203358089281934e-05,
+      "loss": 0.7888,
+      "step": 15370
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.0181614046438165e-05,
+      "loss": 0.7707,
+      "step": 15375
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.0159889253044722e-05,
+      "loss": 0.8008,
+      "step": 15380
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.0138183725951635e-05,
+      "loss": 0.7484,
+      "step": 15385
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.0116497481993986e-05,
+      "loss": 0.7874,
+      "step": 15390
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.0094830537991873e-05,
+      "loss": 0.762,
+      "step": 15395
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.0073182910750465e-05,
+      "loss": 0.7765,
+      "step": 15400
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.0051554617059937e-05,
+      "loss": 0.7788,
+      "step": 15405
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.002994567369545e-05,
+      "loss": 0.7528,
+      "step": 15410
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.0008356097417178e-05,
+      "loss": 0.805,
+      "step": 15415
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 9.986785904970273e-06,
+      "loss": 0.7932,
+      "step": 15420
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 9.965235113084833e-06,
+      "loss": 0.8057,
+      "step": 15425
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 9.94370373847592e-06,
+      "loss": 0.7613,
+      "step": 15430
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 9.92219179784356e-06,
+      "loss": 0.7539,
+      "step": 15435
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 9.900699307872657e-06,
+      "loss": 0.7923,
+      "step": 15440
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 9.879226285233074e-06,
+      "loss": 0.7489,
+      "step": 15445
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 9.857772746579536e-06,
+      "loss": 0.7831,
+      "step": 15450
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 9.836338708551688e-06,
+      "loss": 0.8167,
+      "step": 15455
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 9.814924187774045e-06,
+      "loss": 0.7985,
+      "step": 15460
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 9.793529200855962e-06,
+      "loss": 0.7796,
+      "step": 15465
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 9.772153764391662e-06,
+      "loss": 0.7699,
+      "step": 15470
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 9.750797894960211e-06,
+      "loss": 0.7693,
+      "step": 15475
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 9.729461609125476e-06,
+      "loss": 0.8203,
+      "step": 15480
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 9.708144923436147e-06,
+      "loss": 0.7713,
+      "step": 15485
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 9.686847854425728e-06,
+      "loss": 0.7435,
+      "step": 15490
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 9.66557041861247e-06,
+      "loss": 0.832,
+      "step": 15495
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 9.644312632499425e-06,
+      "loss": 0.7842,
+      "step": 15500
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 9.623074512574405e-06,
+      "loss": 0.7833,
+      "step": 15505
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 9.601856075309967e-06,
+      "loss": 0.7264,
+      "step": 15510
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 9.580657337163386e-06,
+      "loss": 0.7612,
+      "step": 15515
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 9.559478314576648e-06,
+      "loss": 0.827,
+      "step": 15520
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 9.538319023976508e-06,
+      "loss": 0.7636,
+      "step": 15525
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 9.517179481774356e-06,
+      "loss": 0.755,
+      "step": 15530
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 9.496059704366276e-06,
+      "loss": 0.7376,
+      "step": 15535
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 9.474959708133043e-06,
+      "loss": 0.7462,
+      "step": 15540
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 9.453879509440074e-06,
+      "loss": 0.7653,
+      "step": 15545
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 9.432819124637447e-06,
+      "loss": 0.7657,
+      "step": 15550
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 9.411778570059836e-06,
+      "loss": 0.7733,
+      "step": 15555
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 9.390757862026573e-06,
+      "loss": 0.7743,
+      "step": 15560
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 9.369757016841592e-06,
+      "loss": 0.8056,
+      "step": 15565
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 9.348776050793386e-06,
+      "loss": 0.8006,
+      "step": 15570
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 9.327814980155068e-06,
+      "loss": 0.7959,
+      "step": 15575
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 9.306873821184317e-06,
+      "loss": 0.8006,
+      "step": 15580
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 9.285952590123334e-06,
+      "loss": 0.817,
+      "step": 15585
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 9.2650513031989e-06,
+      "loss": 0.8048,
+      "step": 15590
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 9.244169976622328e-06,
+      "loss": 0.8348,
+      "step": 15595
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 9.22330862658941e-06,
+      "loss": 0.789,
+      "step": 15600
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 9.202467269280487e-06,
+      "loss": 0.7707,
+      "step": 15605
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 9.181645920860368e-06,
+      "loss": 0.7519,
+      "step": 15610
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 9.160844597478346e-06,
+      "loss": 0.7741,
+      "step": 15615
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 9.14006331526821e-06,
+      "loss": 0.7837,
+      "step": 15620
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 9.119302090348161e-06,
+      "loss": 0.7914,
+      "step": 15625
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 9.098560938820866e-06,
+      "loss": 0.7799,
+      "step": 15630
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 9.077839876773436e-06,
+      "loss": 0.7258,
+      "step": 15635
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 9.05713892027737e-06,
+      "loss": 0.8028,
+      "step": 15640
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 9.036458085388591e-06,
+      "loss": 0.7634,
+      "step": 15645
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 9.01579738814742e-06,
+      "loss": 0.7848,
+      "step": 15650
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 8.995156844578555e-06,
+      "loss": 0.7523,
+      "step": 15655
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 8.974536470691049e-06,
+      "loss": 0.7779,
+      "step": 15660
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 8.953936282478321e-06,
+      "loss": 0.7584,
+      "step": 15665
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 8.933356295918151e-06,
+      "loss": 0.8097,
+      "step": 15670
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 8.912796526972622e-06,
+      "loss": 0.764,
+      "step": 15675
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 8.892256991588128e-06,
+      "loss": 0.7993,
+      "step": 15680
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 8.87173770569543e-06,
+      "loss": 0.7989,
+      "step": 15685
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 8.851238685209512e-06,
+      "loss": 0.7825,
+      "step": 15690
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 8.830759946029687e-06,
+      "loss": 0.7985,
+      "step": 15695
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 8.810301504039509e-06,
+      "loss": 0.7481,
+      "step": 15700
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 8.789863375106803e-06,
+      "loss": 0.7714,
+      "step": 15705
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 8.769445575083656e-06,
+      "loss": 0.7751,
+      "step": 15710
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 8.749048119806339e-06,
+      "loss": 0.7998,
+      "step": 15715
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 8.72867102509539e-06,
+      "loss": 0.7411,
+      "step": 15720
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 8.708314306755548e-06,
+      "loss": 0.7563,
+      "step": 15725
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 8.687977980575714e-06,
+      "loss": 0.7796,
+      "step": 15730
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 8.667662062329009e-06,
+      "loss": 0.8056,
+      "step": 15735
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 8.647366567772727e-06,
+      "loss": 0.8104,
+      "step": 15740
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 8.62709151264828e-06,
+      "loss": 0.7652,
+      "step": 15745
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 8.606836912681281e-06,
+      "loss": 0.7454,
+      "step": 15750
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 8.586602783581427e-06,
+      "loss": 0.7885,
+      "step": 15755
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 8.566389141042571e-06,
+      "loss": 0.7615,
+      "step": 15760
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 8.546196000742675e-06,
+      "loss": 0.7639,
+      "step": 15765
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 8.526023378343768e-06,
+      "loss": 0.7509,
+      "step": 15770
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 8.505871289492024e-06,
+      "loss": 0.7735,
+      "step": 15775
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 8.485739749817633e-06,
+      "loss": 0.7991,
+      "step": 15780
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 8.465628774934865e-06,
+      "loss": 0.7737,
+      "step": 15785
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 8.445538380442044e-06,
+      "loss": 0.7459,
+      "step": 15790
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 8.425468581921548e-06,
+      "loss": 0.8178,
+      "step": 15795
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 8.405419394939745e-06,
+      "loss": 0.7281,
+      "step": 15800
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 8.385390835047038e-06,
+      "loss": 0.7758,
+      "step": 15805
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 8.365382917777834e-06,
+      "loss": 0.7391,
+      "step": 15810
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 8.345395658650532e-06,
+      "loss": 0.7862,
+      "step": 15815
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 8.325429073167488e-06,
+      "loss": 0.7822,
+      "step": 15820
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 8.305483176815018e-06,
+      "loss": 0.7729,
+      "step": 15825
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 8.285557985063447e-06,
+      "loss": 0.7404,
+      "step": 15830
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 8.265653513366984e-06,
+      "loss": 0.7478,
+      "step": 15835
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 8.245769777163778e-06,
+      "loss": 0.7759,
+      "step": 15840
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 8.225906791875908e-06,
+      "loss": 0.7738,
+      "step": 15845
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 8.206064572909358e-06,
+      "loss": 0.7822,
+      "step": 15850
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 8.186243135654006e-06,
+      "loss": 0.8015,
+      "step": 15855
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 8.166442495483591e-06,
+      "loss": 0.793,
+      "step": 15860
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 8.14666266775575e-06,
+      "loss": 0.7835,
+      "step": 15865
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 8.126903667811968e-06,
+      "loss": 0.7607,
+      "step": 15870
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 8.107165510977552e-06,
+      "loss": 0.7548,
+      "step": 15875
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 8.08744821256168e-06,
+      "loss": 0.7877,
+      "step": 15880
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 8.067751787857337e-06,
+      "loss": 0.8028,
+      "step": 15885
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 8.048076252141301e-06,
+      "loss": 0.7347,
+      "step": 15890
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 8.028421620674173e-06,
+      "loss": 0.8039,
+      "step": 15895
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 8.00878790870034e-06,
+      "loss": 0.7959,
+      "step": 15900
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 7.989175131447936e-06,
+      "loss": 0.7553,
+      "step": 15905
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 7.969583304128897e-06,
+      "loss": 0.7922,
+      "step": 15910
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 7.950012441938864e-06,
+      "loss": 0.7839,
+      "step": 15915
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 7.930462560057257e-06,
+      "loss": 0.8018,
+      "step": 15920
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 7.910933673647215e-06,
+      "loss": 0.7945,
+      "step": 15925
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 7.891425797855568e-06,
+      "loss": 0.7589,
+      "step": 15930
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 7.87193894781288e-06,
+      "loss": 0.7607,
+      "step": 15935
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 7.852473138633406e-06,
+      "loss": 0.7768,
+      "step": 15940
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 7.833028385415045e-06,
+      "loss": 0.7363,
+      "step": 15945
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 7.813604703239408e-06,
+      "loss": 0.7463,
+      "step": 15950
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 7.79420210717175e-06,
+      "loss": 0.7461,
+      "step": 15955
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 7.774820612260949e-06,
+      "loss": 0.841,
+      "step": 15960
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 7.755460233539548e-06,
+      "loss": 0.7678,
+      "step": 15965
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 7.736120986023693e-06,
+      "loss": 0.7791,
+      "step": 15970
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 7.716802884713163e-06,
+      "loss": 0.7916,
+      "step": 15975
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 7.6975059445913e-06,
+      "loss": 0.7489,
+      "step": 15980
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 7.678230180625037e-06,
+      "loss": 0.7699,
+      "step": 15985
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 7.658975607764938e-06,
+      "loss": 0.7367,
+      "step": 15990
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 7.639742240945072e-06,
+      "loss": 0.7671,
+      "step": 15995
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 7.620530095083063e-06,
+      "loss": 0.7528,
+      "step": 16000
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 7.601339185080102e-06,
+      "loss": 0.7727,
+      "step": 16005
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 7.582169525820906e-06,
+      "loss": 0.7781,
+      "step": 16010
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 7.563021132173705e-06,
+      "loss": 0.7268,
+      "step": 16015
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 7.5438940189902145e-06,
+      "loss": 0.7588,
+      "step": 16020
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 7.524788201105673e-06,
+      "loss": 0.7451,
+      "step": 16025
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 7.505703693338802e-06,
+      "loss": 0.7791,
+      "step": 16030
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 7.4866405104917655e-06,
+      "loss": 0.7572,
+      "step": 16035
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 7.467598667350214e-06,
+      "loss": 0.7721,
+      "step": 16040
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 7.448578178683247e-06,
+      "loss": 0.7425,
+      "step": 16045
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 7.429579059243377e-06,
+      "loss": 0.8174,
+      "step": 16050
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 7.410601323766567e-06,
+      "loss": 0.7453,
+      "step": 16055
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 7.391644986972199e-06,
+      "loss": 0.7817,
+      "step": 16060
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 7.372710063563021e-06,
+      "loss": 0.7664,
+      "step": 16065
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 7.353796568225215e-06,
+      "loss": 0.7491,
+      "step": 16070
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 7.334904515628307e-06,
+      "loss": 0.758,
+      "step": 16075
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 7.3160339204252225e-06,
+      "loss": 0.797,
+      "step": 16080
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 7.297184797252231e-06,
+      "loss": 0.7708,
+      "step": 16085
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 7.278357160728938e-06,
+      "loss": 0.7588,
+      "step": 16090
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 7.2595510254582995e-06,
+      "loss": 0.7841,
+      "step": 16095
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 7.240766406026601e-06,
+      "loss": 0.8178,
+      "step": 16100
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 7.222003317003405e-06,
+      "loss": 0.791,
+      "step": 16105
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 7.203261772941608e-06,
+      "loss": 0.741,
+      "step": 16110
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 7.184541788377397e-06,
+      "loss": 0.7568,
+      "step": 16115
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 7.16584337783021e-06,
+      "loss": 0.7987,
+      "step": 16120
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 7.147166555802773e-06,
+      "loss": 0.7606,
+      "step": 16125
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 7.128511336781061e-06,
+      "loss": 0.7916,
+      "step": 16130
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 7.109877735234307e-06,
+      "loss": 0.7425,
+      "step": 16135
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 7.09126576561496e-06,
+      "loss": 0.7623,
+      "step": 16140
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 7.072675442358669e-06,
+      "loss": 0.7972,
+      "step": 16145
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 7.0541067798843685e-06,
+      "loss": 0.8149,
+      "step": 16150
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 7.0355597925941045e-06,
+      "loss": 0.7604,
+      "step": 16155
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 7.0170344948731824e-06,
+      "loss": 0.7563,
+      "step": 16160
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.998530901090033e-06,
+      "loss": 0.7694,
+      "step": 16165
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.980049025596278e-06,
+      "loss": 0.7871,
+      "step": 16170
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.961588882726708e-06,
+      "loss": 0.8013,
+      "step": 16175
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.943150486799219e-06,
+      "loss": 0.7589,
+      "step": 16180
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.9247338521148724e-06,
+      "loss": 0.7552,
+      "step": 16185
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.906338992957845e-06,
+      "loss": 0.7878,
+      "step": 16190
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.887965923595405e-06,
+      "loss": 0.7836,
+      "step": 16195
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.8696146582779385e-06,
+      "loss": 0.7716,
+      "step": 16200
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.851285211238936e-06,
+      "loss": 0.7515,
+      "step": 16205
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.832977596694918e-06,
+      "loss": 0.7557,
+      "step": 16210
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.814691828845508e-06,
+      "loss": 0.7673,
+      "step": 16215
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.7964279218733916e-06,
+      "loss": 0.7734,
+      "step": 16220
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.7781858899442636e-06,
+      "loss": 0.7875,
+      "step": 16225
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.759965747206883e-06,
+      "loss": 0.7848,
+      "step": 16230
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.741767507793002e-06,
+      "loss": 0.7715,
+      "step": 16235
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.723591185817437e-06,
+      "loss": 0.7524,
+      "step": 16240
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.705436795377946e-06,
+      "loss": 0.7847,
+      "step": 16245
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.687304350555301e-06,
+      "loss": 0.7592,
+      "step": 16250
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.669193865413255e-06,
+      "loss": 0.7701,
+      "step": 16255
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.651105353998538e-06,
+      "loss": 0.8025,
+      "step": 16260
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.633038830340807e-06,
+      "loss": 0.788,
+      "step": 16265
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.614994308452696e-06,
+      "loss": 0.7704,
+      "step": 16270
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.596971802329756e-06,
+      "loss": 0.8066,
+      "step": 16275
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.578971325950481e-06,
+      "loss": 0.8085,
+      "step": 16280
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.5609928932762555e-06,
+      "loss": 0.7834,
+      "step": 16285
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.5430365182513574e-06,
+      "loss": 0.7716,
+      "step": 16290
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.525102214803015e-06,
+      "loss": 0.8118,
+      "step": 16295
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.507189996841276e-06,
+      "loss": 0.7915,
+      "step": 16300
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.489299878259066e-06,
+      "loss": 0.79,
+      "step": 16305
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.471431872932218e-06,
+      "loss": 0.7786,
+      "step": 16310
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.453585994719352e-06,
+      "loss": 0.7918,
+      "step": 16315
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.435762257461977e-06,
+      "loss": 0.7362,
+      "step": 16320
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.417960674984385e-06,
+      "loss": 0.7347,
+      "step": 16325
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.4001812610937156e-06,
+      "loss": 0.7441,
+      "step": 16330
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.382424029579912e-06,
+      "loss": 0.781,
+      "step": 16335
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.364688994215692e-06,
+      "loss": 0.758,
+      "step": 16340
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.346976168756572e-06,
+      "loss": 0.7874,
+      "step": 16345
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.329285566940856e-06,
+      "loss": 0.8125,
+      "step": 16350
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.311617202489576e-06,
+      "loss": 0.7481,
+      "step": 16355
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.2939710891065405e-06,
+      "loss": 0.7109,
+      "step": 16360
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.276347240478313e-06,
+      "loss": 0.8011,
+      "step": 16365
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.25874567027415e-06,
+      "loss": 0.7979,
+      "step": 16370
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.241166392146063e-06,
+      "loss": 0.7923,
+      "step": 16375
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.223609419728745e-06,
+      "loss": 0.8054,
+      "step": 16380
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.206074766639614e-06,
+      "loss": 0.8137,
+      "step": 16385
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.188562446478771e-06,
+      "loss": 0.7723,
+      "step": 16390
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.1710724728289685e-06,
+      "loss": 0.7971,
+      "step": 16395
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.153604859255681e-06,
+      "loss": 0.7445,
+      "step": 16400
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.136159619306989e-06,
+      "loss": 0.7571,
+      "step": 16405
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.118736766513632e-06,
+      "loss": 0.8123,
+      "step": 16410
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.101336314389001e-06,
+      "loss": 0.7688,
+      "step": 16415
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.083958276429114e-06,
+      "loss": 0.7843,
+      "step": 16420
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.066602666112576e-06,
+      "loss": 0.7496,
+      "step": 16425
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.0492694969006215e-06,
+      "loss": 0.7245,
+      "step": 16430
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.031958782237077e-06,
+      "loss": 0.7675,
+      "step": 16435
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.014670535548354e-06,
+      "loss": 0.7545,
+      "step": 16440
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.997404770243426e-06,
+      "loss": 0.7952,
+      "step": 16445
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.9801614997138185e-06,
+      "loss": 0.7744,
+      "step": 16450
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.962940737333664e-06,
+      "loss": 0.7703,
+      "step": 16455
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.945742496459584e-06,
+      "loss": 0.7981,
+      "step": 16460
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.928566790430739e-06,
+      "loss": 0.8065,
+      "step": 16465
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.911413632568829e-06,
+      "loss": 0.7922,
+      "step": 16470
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.894283036178057e-06,
+      "loss": 0.7841,
+      "step": 16475
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.877175014545141e-06,
+      "loss": 0.8094,
+      "step": 16480
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.860089580939252e-06,
+      "loss": 0.7487,
+      "step": 16485
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.8430267486120845e-06,
+      "loss": 0.7645,
+      "step": 16490
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.825986530797783e-06,
+      "loss": 0.7306,
+      "step": 16495
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.808968940712939e-06,
+      "loss": 0.7521,
+      "step": 16500
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.791973991556618e-06,
+      "loss": 0.7549,
+      "step": 16505
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.775001696510328e-06,
+      "loss": 0.6945,
+      "step": 16510
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.758052068737966e-06,
+      "loss": 0.8001,
+      "step": 16515
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.741125121385894e-06,
+      "loss": 0.7907,
+      "step": 16520
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.724220867582866e-06,
+      "loss": 0.7723,
+      "step": 16525
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.707339320440016e-06,
+      "loss": 0.7614,
+      "step": 16530
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.690480493050899e-06,
+      "loss": 0.7901,
+      "step": 16535
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.673644398491421e-06,
+      "loss": 0.8193,
+      "step": 16540
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.656831049819873e-06,
+      "loss": 0.7807,
+      "step": 16545
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.640040460076901e-06,
+      "loss": 0.7512,
+      "step": 16550
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.6232726422854904e-06,
+      "loss": 0.7837,
+      "step": 16555
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.606527609450968e-06,
+      "loss": 0.7501,
+      "step": 16560
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.589805374561007e-06,
+      "loss": 0.7688,
+      "step": 16565
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.573105950585559e-06,
+      "loss": 0.7987,
+      "step": 16570
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.556429350476916e-06,
+      "loss": 0.7472,
+      "step": 16575
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.5397755871696695e-06,
+      "loss": 0.7648,
+      "step": 16580
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.523144673580666e-06,
+      "loss": 0.791,
+      "step": 16585
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.506536622609059e-06,
+      "loss": 0.7904,
+      "step": 16590
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.489951447136264e-06,
+      "loss": 0.809,
+      "step": 16595
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.473389160025951e-06,
+      "loss": 0.8065,
+      "step": 16600
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.456849774124035e-06,
+      "loss": 0.7966,
+      "step": 16605
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.440333302258649e-06,
+      "loss": 0.767,
+      "step": 16610
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.423839757240215e-06,
+      "loss": 0.7412,
+      "step": 16615
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.4073691518613e-06,
+      "loss": 0.77,
+      "step": 16620
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.3909214988967345e-06,
+      "loss": 0.7921,
+      "step": 16625
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.374496811103502e-06,
+      "loss": 0.7587,
+      "step": 16630
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.3580951012208066e-06,
+      "loss": 0.7757,
+      "step": 16635
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.341716381970031e-06,
+      "loss": 0.7611,
+      "step": 16640
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.325360666054691e-06,
+      "loss": 0.7531,
+      "step": 16645
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.3090279661604945e-06,
+      "loss": 0.7268,
+      "step": 16650
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.2927182949553e-06,
+      "loss": 0.7999,
+      "step": 16655
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.276431665089074e-06,
+      "loss": 0.7889,
+      "step": 16660
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.260168089193935e-06,
+      "loss": 0.7692,
+      "step": 16665
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.243927579884129e-06,
+      "loss": 0.7518,
+      "step": 16670
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.227710149755976e-06,
+      "loss": 0.7214,
+      "step": 16675
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.211515811387933e-06,
+      "loss": 0.7579,
+      "step": 16680
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.195344577340531e-06,
+      "loss": 0.7704,
+      "step": 16685
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.179196460156371e-06,
+      "loss": 0.7583,
+      "step": 16690
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.163071472360152e-06,
+      "loss": 0.7772,
+      "step": 16695
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.146969626458584e-06,
+      "loss": 0.7473,
+      "step": 16700
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.130890934940501e-06,
+      "loss": 0.7587,
+      "step": 16705
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.114835410276712e-06,
+      "loss": 0.7588,
+      "step": 16710
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.098803064920085e-06,
+      "loss": 0.772,
+      "step": 16715
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.082793911305505e-06,
+      "loss": 0.7943,
+      "step": 16720
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.0668079618498865e-06,
+      "loss": 0.7802,
+      "step": 16725
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.050845228952116e-06,
+      "loss": 0.7868,
+      "step": 16730
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.034905724993093e-06,
+      "loss": 0.763,
+      "step": 16735
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.018989462335696e-06,
+      "loss": 0.7764,
+      "step": 16740
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.003096453324788e-06,
+      "loss": 0.7553,
+      "step": 16745
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.987226710287167e-06,
+      "loss": 0.7703,
+      "step": 16750
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.971380245531611e-06,
+      "loss": 0.7638,
+      "step": 16755
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.955557071348852e-06,
+      "loss": 0.7683,
+      "step": 16760
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.939757200011523e-06,
+      "loss": 0.7558,
+      "step": 16765
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.9239806437741915e-06,
+      "loss": 0.8,
+      "step": 16770
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.908227414873383e-06,
+      "loss": 0.7855,
+      "step": 16775
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.892497525527473e-06,
+      "loss": 0.7533,
+      "step": 16780
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.87679098793678e-06,
+      "loss": 0.7576,
+      "step": 16785
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.861107814283474e-06,
+      "loss": 0.781,
+      "step": 16790
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.8454480167316255e-06,
+      "loss": 0.7828,
+      "step": 16795
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.829811607427178e-06,
+      "loss": 0.7186,
+      "step": 16800
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.81419859849791e-06,
+      "loss": 0.8092,
+      "step": 16805
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.798609002053476e-06,
+      "loss": 0.7764,
+      "step": 16810
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.783042830185372e-06,
+      "loss": 0.7739,
+      "step": 16815
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.7675000949668926e-06,
+      "loss": 0.7711,
+      "step": 16820
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.75198080845319e-06,
+      "loss": 0.7903,
+      "step": 16825
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.736484982681225e-06,
+      "loss": 0.7539,
+      "step": 16830
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.721012629669734e-06,
+      "loss": 0.7986,
+      "step": 16835
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.705563761419284e-06,
+      "loss": 0.7898,
+      "step": 16840
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.6901383899122075e-06,
+      "loss": 0.7789,
+      "step": 16845
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.674736527112605e-06,
+      "loss": 0.7942,
+      "step": 16850
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.659358184966372e-06,
+      "loss": 0.7595,
+      "step": 16855
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.644003375401115e-06,
+      "loss": 0.7734,
+      "step": 16860
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.628672110326245e-06,
+      "loss": 0.7602,
+      "step": 16865
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.613364401632874e-06,
+      "loss": 0.7831,
+      "step": 16870
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.598080261193845e-06,
+      "loss": 0.7192,
+      "step": 16875
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.582819700863728e-06,
+      "loss": 0.7833,
+      "step": 16880
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.5675827324788164e-06,
+      "loss": 0.7839,
+      "step": 16885
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.55236936785708e-06,
+      "loss": 0.7423,
+      "step": 16890
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.5371796187982e-06,
+      "loss": 0.7823,
+      "step": 16895
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.522013497083539e-06,
+      "loss": 0.7759,
+      "step": 16900
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.50687101447613e-06,
+      "loss": 0.8036,
+      "step": 16905
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.491752182720672e-06,
+      "loss": 0.7691,
+      "step": 16910
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.476657013543503e-06,
+      "loss": 0.7716,
+      "step": 16915
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.461585518652653e-06,
+      "loss": 0.7721,
+      "step": 16920
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.446537709737745e-06,
+      "loss": 0.7696,
+      "step": 16925
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.431513598470039e-06,
+      "loss": 0.776,
+      "step": 16930
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.416513196502451e-06,
+      "loss": 0.7877,
+      "step": 16935
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.401536515469453e-06,
+      "loss": 0.7681,
+      "step": 16940
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.386583566987169e-06,
+      "loss": 0.7918,
+      "step": 16945
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.371654362653269e-06,
+      "loss": 0.7915,
+      "step": 16950
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.356748914047043e-06,
+      "loss": 0.7538,
+      "step": 16955
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.341867232729354e-06,
+      "loss": 0.742,
+      "step": 16960
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.327009330242597e-06,
+      "loss": 0.7547,
+      "step": 16965
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.31217521811076e-06,
+      "loss": 0.7883,
+      "step": 16970
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.297364907839372e-06,
+      "loss": 0.7878,
+      "step": 16975
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.2825784109154805e-06,
+      "loss": 0.7677,
+      "step": 16980
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.2678157388076825e-06,
+      "loss": 0.757,
+      "step": 16985
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.253076902966102e-06,
+      "loss": 0.7571,
+      "step": 16990
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.238361914822347e-06,
+      "loss": 0.8023,
+      "step": 16995
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.2236707857895585e-06,
+      "loss": 0.7954,
+      "step": 17000
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.20900352726235e-06,
+      "loss": 0.7443,
+      "step": 17005
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.194360150616834e-06,
+      "loss": 0.7712,
+      "step": 17010
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.1797406672106065e-06,
+      "loss": 0.7176,
+      "step": 17015
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.165145088382706e-06,
+      "loss": 0.7709,
+      "step": 17020
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.150573425453651e-06,
+      "loss": 0.7736,
+      "step": 17025
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.136025689725415e-06,
+      "loss": 0.8031,
+      "step": 17030
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.121501892481392e-06,
+      "loss": 0.7647,
+      "step": 17035
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.107002044986421e-06,
+      "loss": 0.7687,
+      "step": 17040
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.092526158486779e-06,
+      "loss": 0.7992,
+      "step": 17045
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.078074244210125e-06,
+      "loss": 0.7885,
+      "step": 17050
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.063646313365554e-06,
+      "loss": 0.7702,
+      "step": 17055
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.0492423771435475e-06,
+      "loss": 0.7788,
+      "step": 17060
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.034862446715988e-06,
+      "loss": 0.7337,
+      "step": 17065
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.020506533236125e-06,
+      "loss": 0.7736,
+      "step": 17070
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.006174647838563e-06,
+      "loss": 0.7601,
+      "step": 17075
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.991866801639321e-06,
+      "loss": 0.7842,
+      "step": 17080
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.977583005735726e-06,
+      "loss": 0.8225,
+      "step": 17085
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.963323271206484e-06,
+      "loss": 0.76,
+      "step": 17090
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.949087609111603e-06,
+      "loss": 0.7553,
+      "step": 17095
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.934876030492454e-06,
+      "loss": 0.7673,
+      "step": 17100
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.920688546371715e-06,
+      "loss": 0.7626,
+      "step": 17105
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.906525167753368e-06,
+      "loss": 0.824,
+      "step": 17110
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.89238590562271e-06,
+      "loss": 0.7848,
+      "step": 17115
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.8782707709463385e-06,
+      "loss": 0.7337,
+      "step": 17120
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.864179774672113e-06,
+      "loss": 0.7276,
+      "step": 17125
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.850112927729193e-06,
+      "loss": 0.7439,
+      "step": 17130
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.836070241028009e-06,
+      "loss": 0.7823,
+      "step": 17135
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.8220517254602314e-06,
+      "loss": 0.8117,
+      "step": 17140
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.808057391898801e-06,
+      "loss": 0.7862,
+      "step": 17145
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.794087251197905e-06,
+      "loss": 0.7705,
+      "step": 17150
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.780141314192944e-06,
+      "loss": 0.7703,
+      "step": 17155
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.7662195917005806e-06,
+      "loss": 0.8021,
+      "step": 17160
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.7523220945186475e-06,
+      "loss": 0.7899,
+      "step": 17165
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.7384488334262508e-06,
+      "loss": 0.833,
+      "step": 17170
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.724599819183652e-06,
+      "loss": 0.7689,
+      "step": 17175
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.7107750625323078e-06,
+      "loss": 0.8011,
+      "step": 17180
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.696974574194885e-06,
+      "loss": 0.8219,
+      "step": 17185
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.6831983648752177e-06,
+      "loss": 0.8026,
+      "step": 17190
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.6694464452582927e-06,
+      "loss": 0.7813,
+      "step": 17195
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.655718826010275e-06,
+      "loss": 0.7536,
+      "step": 17200
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.6420155177784867e-06,
+      "loss": 0.7805,
+      "step": 17205
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.628336531191381e-06,
+      "loss": 0.7847,
+      "step": 17210
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.6146818768585456e-06,
+      "loss": 0.7962,
+      "step": 17215
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.6010515653707028e-06,
+      "loss": 0.7891,
+      "step": 17220
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.5874456072996977e-06,
+      "loss": 0.7757,
+      "step": 17225
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.5738640131984827e-06,
+      "loss": 0.7032,
+      "step": 17230
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.5603067936010897e-06,
+      "loss": 0.764,
+      "step": 17235
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.5467739590227023e-06,
+      "loss": 0.7587,
+      "step": 17240
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.5332655199595283e-06,
+      "loss": 0.7708,
+      "step": 17245
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.5197814868889028e-06,
+      "loss": 0.7403,
+      "step": 17250
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.506321870269191e-06,
+      "loss": 0.7453,
+      "step": 17255
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.492886680539849e-06,
+      "loss": 0.7823,
+      "step": 17260
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.4794759281213856e-06,
+      "loss": 0.7588,
+      "step": 17265
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.4660896234153336e-06,
+      "loss": 0.7965,
+      "step": 17270
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.4527277768042798e-06,
+      "loss": 0.8028,
+      "step": 17275
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.4393903986518547e-06,
+      "loss": 0.7676,
+      "step": 17280
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.426077499302681e-06,
+      "loss": 0.7366,
+      "step": 17285
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.41278908908241e-06,
+      "loss": 0.7944,
+      "step": 17290
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.3995251782977082e-06,
+      "loss": 0.7684,
+      "step": 17295
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.3862857772362236e-06,
+      "loss": 0.7534,
+      "step": 17300
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.3730708961665964e-06,
+      "loss": 0.7965,
+      "step": 17305
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.3598805453384712e-06,
+      "loss": 0.7372,
+      "step": 17310
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.3467147349824286e-06,
+      "loss": 0.7323,
+      "step": 17315
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.3335734753100507e-06,
+      "loss": 0.7821,
+      "step": 17320
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.320456776513845e-06,
+      "loss": 0.7891,
+      "step": 17325
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.30736464876731e-06,
+      "loss": 0.8135,
+      "step": 17330
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.2942971022248528e-06,
+      "loss": 0.7467,
+      "step": 17335
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.2812541470218166e-06,
+      "loss": 0.7981,
+      "step": 17340
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.268235793274494e-06,
+      "loss": 0.8145,
+      "step": 17345
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.2552420510800814e-06,
+      "loss": 0.8238,
+      "step": 17350
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.242272930516677e-06,
+      "loss": 0.7708,
+      "step": 17355
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.2293284416433006e-06,
+      "loss": 0.7592,
+      "step": 17360
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.216408594499857e-06,
+      "loss": 0.7501,
+      "step": 17365
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.203513399107152e-06,
+      "loss": 0.772,
+      "step": 17370
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.190642865466844e-06,
+      "loss": 0.7794,
+      "step": 17375
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.1777970035614864e-06,
+      "loss": 0.77,
+      "step": 17380
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.1649758233544975e-06,
+      "loss": 0.7663,
+      "step": 17385
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.1521793347901414e-06,
+      "loss": 0.7894,
+      "step": 17390
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.139407547793514e-06,
+      "loss": 0.7856,
+      "step": 17395
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.126660472270608e-06,
+      "loss": 0.7514,
+      "step": 17400
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.1139381181081864e-06,
+      "loss": 0.7228,
+      "step": 17405
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.101240495173886e-06,
+      "loss": 0.7886,
+      "step": 17410
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.0885676133161234e-06,
+      "loss": 0.8148,
+      "step": 17415
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.0759194823641532e-06,
+      "loss": 0.791,
+      "step": 17420
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.0632961121280314e-06,
+      "loss": 0.7551,
+      "step": 17425
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.050697512398589e-06,
+      "loss": 0.7729,
+      "step": 17430
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.03812369294746e-06,
+      "loss": 0.7871,
+      "step": 17435
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.0255746635270683e-06,
+      "loss": 0.765,
+      "step": 17440
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.0130504338705793e-06,
+      "loss": 0.7961,
+      "step": 17445
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.0005510136919503e-06,
+      "loss": 0.8163,
+      "step": 17450
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.9880764126858974e-06,
+      "loss": 0.815,
+      "step": 17455
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.975626640527859e-06,
+      "loss": 0.7801,
+      "step": 17460
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.963201706874042e-06,
+      "loss": 0.7969,
+      "step": 17465
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.9508016213613926e-06,
+      "loss": 0.7622,
+      "step": 17470
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.938426393607548e-06,
+      "loss": 0.7982,
+      "step": 17475
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.9260760332109146e-06,
+      "loss": 0.7945,
+      "step": 17480
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.9137505497505593e-06,
+      "loss": 0.7602,
+      "step": 17485
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.901449952786307e-06,
+      "loss": 0.756,
+      "step": 17490
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.889174251858642e-06,
+      "loss": 0.8209,
+      "step": 17495
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.8769234564887493e-06,
+      "loss": 0.7736,
+      "step": 17500
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.864697576178501e-06,
+      "loss": 0.8032,
+      "step": 17505
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.8524966204104495e-06,
+      "loss": 0.8096,
+      "step": 17510
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.840320598647798e-06,
+      "loss": 0.7593,
+      "step": 17515
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.828169520334425e-06,
+      "loss": 0.7363,
+      "step": 17520
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.816043394894863e-06,
+      "loss": 0.7801,
+      "step": 17525
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.8039422317342864e-06,
+      "loss": 0.7497,
+      "step": 17530
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.7918660402385066e-06,
+      "loss": 0.7771,
+      "step": 17535
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.7798148297739534e-06,
+      "loss": 0.8185,
+      "step": 17540
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.7677886096877243e-06,
+      "loss": 0.7553,
+      "step": 17545
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.755787389307484e-06,
+      "loss": 0.7581,
+      "step": 17550
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.7438111779415454e-06,
+      "loss": 0.7585,
+      "step": 17555
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.7318599848787905e-06,
+      "loss": 0.8115,
+      "step": 17560
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.7199338193887246e-06,
+      "loss": 0.7621,
+      "step": 17565
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.7080326907214327e-06,
+      "loss": 0.7574,
+      "step": 17570
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.6961566081075718e-06,
+      "loss": 0.7671,
+      "step": 17575
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.684305580758385e-06,
+      "loss": 0.7493,
+      "step": 17580
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.6724796178656773e-06,
+      "loss": 0.7647,
+      "step": 17585
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.6606787286018073e-06,
+      "loss": 0.7903,
+      "step": 17590
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.6489029221196963e-06,
+      "loss": 0.7773,
+      "step": 17595
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.637152207552813e-06,
+      "loss": 0.7803,
+      "step": 17600
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.625426594015147e-06,
+      "loss": 0.7386,
+      "step": 17605
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.613726090601236e-06,
+      "loss": 0.7616,
+      "step": 17610
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.602050706386145e-06,
+      "loss": 0.7604,
+      "step": 17615
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.5904004504254337e-06,
+      "loss": 0.7656,
+      "step": 17620
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.5787753317551974e-06,
+      "loss": 0.7314,
+      "step": 17625
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.5671753593920046e-06,
+      "loss": 0.773,
+      "step": 17630
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.555600542332965e-06,
+      "loss": 0.7543,
+      "step": 17635
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.544050889555642e-06,
+      "loss": 0.801,
+      "step": 17640
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.5325264100180814e-06,
+      "loss": 0.7588,
+      "step": 17645
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.5210271126588177e-06,
+      "loss": 0.7607,
+      "step": 17650
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.50955300639686e-06,
+      "loss": 0.8088,
+      "step": 17655
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.498104100131652e-06,
+      "loss": 0.7906,
+      "step": 17660
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.48668040274312e-06,
+      "loss": 0.7713,
+      "step": 17665
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.4752819230916198e-06,
+      "loss": 0.7606,
+      "step": 17670
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.463908670017968e-06,
+      "loss": 0.7695,
+      "step": 17675
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.4525606523433825e-06,
+      "loss": 0.7708,
+      "step": 17680
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.4412378788695386e-06,
+      "loss": 0.7945,
+      "step": 17685
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.4299403583785287e-06,
+      "loss": 0.7441,
+      "step": 17690
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.418668099632839e-06,
+      "loss": 0.791,
+      "step": 17695
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.4074211113753627e-06,
+      "loss": 0.7672,
+      "step": 17700
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.396199402329433e-06,
+      "loss": 0.7836,
+      "step": 17705
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3850029811987276e-06,
+      "loss": 0.7996,
+      "step": 17710
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.373831856667341e-06,
+      "loss": 0.7752,
+      "step": 17715
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.362686037399728e-06,
+      "loss": 0.7966,
+      "step": 17720
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.351565532040733e-06,
+      "loss": 0.7699,
+      "step": 17725
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.340470349215556e-06,
+      "loss": 0.7356,
+      "step": 17730
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.329400497529757e-06,
+      "loss": 0.7573,
+      "step": 17735
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3183559855692506e-06,
+      "loss": 0.7608,
+      "step": 17740
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.307336821900311e-06,
+      "loss": 0.7688,
+      "step": 17745
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.2963430150695213e-06,
+      "loss": 0.7703,
+      "step": 17750
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.2853745736038233e-06,
+      "loss": 0.7871,
+      "step": 17755
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.2744315060104845e-06,
+      "loss": 0.7923,
+      "step": 17760
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.263513820777072e-06,
+      "loss": 0.7646,
+      "step": 17765
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.2526215263714864e-06,
+      "loss": 0.7614,
+      "step": 17770
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.2417546312419276e-06,
+      "loss": 0.8022,
+      "step": 17775
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.230913143816891e-06,
+      "loss": 0.8126,
+      "step": 17780
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.220097072505178e-06,
+      "loss": 0.7974,
+      "step": 17785
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.2093064256958516e-06,
+      "loss": 0.7839,
+      "step": 17790
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.1985412117582955e-06,
+      "loss": 0.8166,
+      "step": 17795
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.187801439042132e-06,
+      "loss": 0.7879,
+      "step": 17800
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.1770871158772583e-06,
+      "loss": 0.83,
+      "step": 17805
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.1663982505738398e-06,
+      "loss": 0.8207,
+      "step": 17810
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.1557348514223055e-06,
+      "loss": 0.796,
+      "step": 17815
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.1450969266933067e-06,
+      "loss": 0.7987,
+      "step": 17820
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.134484484637758e-06,
+      "loss": 0.741,
+      "step": 17825
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.1238975334867957e-06,
+      "loss": 0.7709,
+      "step": 17830
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.1133360814518053e-06,
+      "loss": 0.7646,
+      "step": 17835
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.1028001367243653e-06,
+      "loss": 0.7731,
+      "step": 17840
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.092289707476296e-06,
+      "loss": 0.7955,
+      "step": 17845
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.0818048018596144e-06,
+      "loss": 0.7922,
+      "step": 17850
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.0713454280065455e-06,
+      "loss": 0.7577,
+      "step": 17855
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.060911594029503e-06,
+      "loss": 0.7902,
+      "step": 17860
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.0505033080211142e-06,
+      "loss": 0.7475,
+      "step": 17865
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.0401205780541585e-06,
+      "loss": 0.7704,
+      "step": 17870
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.029763412181625e-06,
+      "loss": 0.7804,
+      "step": 17875
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.0194318184366532e-06,
+      "loss": 0.7845,
+      "step": 17880
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.009125804832559e-06,
+      "loss": 0.7891,
+      "step": 17885
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9988453793628158e-06,
+      "loss": 0.7439,
+      "step": 17890
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.988590550001046e-06,
+      "loss": 0.7864,
+      "step": 17895
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.978361324701026e-06,
+      "loss": 0.7832,
+      "step": 17900
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9681577113966775e-06,
+      "loss": 0.7533,
+      "step": 17905
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9579797180020406e-06,
+      "loss": 0.7916,
+      "step": 17910
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.947827352411293e-06,
+      "loss": 0.7581,
+      "step": 17915
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9377006224987512e-06,
+      "loss": 0.7618,
+      "step": 17920
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9275995361188175e-06,
+      "loss": 0.8046,
+      "step": 17925
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9175241011060256e-06,
+      "loss": 0.7966,
+      "step": 17930
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9074743252750145e-06,
+      "loss": 0.7574,
+      "step": 17935
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8974502164205092e-06,
+      "loss": 0.7872,
+      "step": 17940
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8874517823173379e-06,
+      "loss": 0.7474,
+      "step": 17945
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8774790307204002e-06,
+      "loss": 0.7644,
+      "step": 17950
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8675319693647073e-06,
+      "loss": 0.7654,
+      "step": 17955
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8576106059653133e-06,
+      "loss": 0.7634,
+      "step": 17960
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8477149482173426e-06,
+      "loss": 0.7799,
+      "step": 17965
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8378450037960042e-06,
+      "loss": 0.7736,
+      "step": 17970
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.828000780356546e-06,
+      "loss": 0.7435,
+      "step": 17975
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.818182285534271e-06,
+      "loss": 0.7809,
+      "step": 17980
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8083895269445237e-06,
+      "loss": 0.7506,
+      "step": 17985
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.798622512182692e-06,
+      "loss": 0.766,
+      "step": 17990
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.7888812488241983e-06,
+      "loss": 0.7746,
+      "step": 17995
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.7791657444244805e-06,
+      "loss": 0.7802,
+      "step": 18000
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7694760065190118e-06,
+      "loss": 0.7611,
+      "step": 18005
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.759812042623276e-06,
+      "loss": 0.7746,
+      "step": 18010
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7501738602327613e-06,
+      "loss": 0.735,
+      "step": 18015
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7405614668229483e-06,
+      "loss": 0.7282,
+      "step": 18020
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7309748698493558e-06,
+      "loss": 0.8027,
+      "step": 18025
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7214140767474483e-06,
+      "loss": 0.7783,
+      "step": 18030
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7118790949327055e-06,
+      "loss": 0.8052,
+      "step": 18035
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7023699318005762e-06,
+      "loss": 0.8132,
+      "step": 18040
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.692886594726482e-06,
+      "loss": 0.8113,
+      "step": 18045
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.6834290910658282e-06,
+      "loss": 0.7893,
+      "step": 18050
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.6739974281539626e-06,
+      "loss": 0.721,
+      "step": 18055
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.664591613306201e-06,
+      "loss": 0.7456,
+      "step": 18060
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.6552116538178206e-06,
+      "loss": 0.7333,
+      "step": 18065
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.6458575569640247e-06,
+      "loss": 0.7335,
+      "step": 18070
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.6365293299999738e-06,
+      "loss": 0.7799,
+      "step": 18075
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.6272269801607579e-06,
+      "loss": 0.7598,
+      "step": 18080
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.6179505146613897e-06,
+      "loss": 0.8198,
+      "step": 18085
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.6086999406968154e-06,
+      "loss": 0.8047,
+      "step": 18090
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.5994752654418957e-06,
+      "loss": 0.7858,
+      "step": 18095
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.5902764960514097e-06,
+      "loss": 0.7674,
+      "step": 18100
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.5811036396600313e-06,
+      "loss": 0.7625,
+      "step": 18105
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.5719567033823332e-06,
+      "loss": 0.7861,
+      "step": 18110
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.5628356943128184e-06,
+      "loss": 0.7441,
+      "step": 18115
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.5537406195258377e-06,
+      "loss": 0.8019,
+      "step": 18120
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.5446714860756414e-06,
+      "loss": 0.7413,
+      "step": 18125
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.5356283009963743e-06,
+      "loss": 0.7781,
+      "step": 18130
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.5266110713020375e-06,
+      "loss": 0.7496,
+      "step": 18135
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.5176198039865111e-06,
+      "loss": 0.803,
+      "step": 18140
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.5086545060235282e-06,
+      "loss": 0.7651,
+      "step": 18145
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.4997151843666893e-06,
+      "loss": 0.746,
+      "step": 18150
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.4908018459494509e-06,
+      "loss": 0.7688,
+      "step": 18155
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.4819144976851022e-06,
+      "loss": 0.7904,
+      "step": 18160
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.4730531464667734e-06,
+      "loss": 0.7649,
+      "step": 18165
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.4642177991674582e-06,
+      "loss": 0.7797,
+      "step": 18170
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.455408462639948e-06,
+      "loss": 0.7481,
+      "step": 18175
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.4466251437168797e-06,
+      "loss": 0.7605,
+      "step": 18180
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.437867849210706e-06,
+      "loss": 0.7933,
+      "step": 18185
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.429136585913686e-06,
+      "loss": 0.7472,
+      "step": 18190
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.4204313605979117e-06,
+      "loss": 0.7699,
+      "step": 18195
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.4117521800152493e-06,
+      "loss": 0.7984,
+      "step": 18200
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.40309905089739e-06,
+      "loss": 0.7494,
+      "step": 18205
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.3944719799558075e-06,
+      "loss": 0.7584,
+      "step": 18210
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.3858709738817575e-06,
+      "loss": 0.7775,
+      "step": 18215
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.3772960393462978e-06,
+      "loss": 0.7639,
+      "step": 18220
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.3687471830002527e-06,
+      "loss": 0.7535,
+      "step": 18225
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.3602244114742172e-06,
+      "loss": 0.7706,
+      "step": 18230
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.3517277313785606e-06,
+      "loss": 0.7582,
+      "step": 18235
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.3432571493034195e-06,
+      "loss": 0.7925,
+      "step": 18240
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.3348126718186809e-06,
+      "loss": 0.7679,
+      "step": 18245
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.3263943054739834e-06,
+      "loss": 0.7729,
+      "step": 18250
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.318002056798717e-06,
+      "loss": 0.7968,
+      "step": 18255
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.3096359323020228e-06,
+      "loss": 0.7742,
+      "step": 18260
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.3012959384727655e-06,
+      "loss": 0.8055,
+      "step": 18265
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.292982081779546e-06,
+      "loss": 0.7736,
+      "step": 18270
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.2846943686707007e-06,
+      "loss": 0.7572,
+      "step": 18275
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.2764328055742818e-06,
+      "loss": 0.7579,
+      "step": 18280
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.268197398898058e-06,
+      "loss": 0.7725,
+      "step": 18285
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.2599881550295188e-06,
+      "loss": 0.8161,
+      "step": 18290
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.2518050803358492e-06,
+      "loss": 0.8053,
+      "step": 18295
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.2436481811639553e-06,
+      "loss": 0.7511,
+      "step": 18300
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.2355174638404204e-06,
+      "loss": 0.7557,
+      "step": 18305
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.2274129346715318e-06,
+      "loss": 0.7603,
+      "step": 18310
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.2193345999432662e-06,
+      "loss": 0.7546,
+      "step": 18315
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.2112824659212862e-06,
+      "loss": 0.7495,
+      "step": 18320
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.2032565388509107e-06,
+      "loss": 0.7869,
+      "step": 18325
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.1952568249571647e-06,
+      "loss": 0.7737,
+      "step": 18330
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.1872833304447227e-06,
+      "loss": 0.7405,
+      "step": 18335
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.1793360614979246e-06,
+      "loss": 0.7458,
+      "step": 18340
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.171415024280773e-06,
+      "loss": 0.7892,
+      "step": 18345
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.1635202249369197e-06,
+      "loss": 0.7522,
+      "step": 18350
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.1556516695896788e-06,
+      "loss": 0.754,
+      "step": 18355
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.1478093643419878e-06,
+      "loss": 0.7739,
+      "step": 18360
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.1399933152764412e-06,
+      "loss": 0.749,
+      "step": 18365
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.1322035284552728e-06,
+      "loss": 0.7673,
+      "step": 18370
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.1244400099203311e-06,
+      "loss": 0.7757,
+      "step": 18375
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.1167027656930955e-06,
+      "loss": 0.7843,
+      "step": 18380
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.1089918017746763e-06,
+      "loss": 0.796,
+      "step": 18385
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.1013071241457832e-06,
+      "loss": 0.7518,
+      "step": 18390
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.0936487387667609e-06,
+      "loss": 0.8087,
+      "step": 18395
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.0860166515775414e-06,
+      "loss": 0.758,
+      "step": 18400
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.0784108684976689e-06,
+      "loss": 0.7645,
+      "step": 18405
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.0708313954262865e-06,
+      "loss": 0.7436,
+      "step": 18410
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.063278238242114e-06,
+      "loss": 0.7836,
+      "step": 18415
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.055751402803498e-06,
+      "loss": 0.7862,
+      "step": 18420
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.0482508949483304e-06,
+      "loss": 0.7922,
+      "step": 18425
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.0407767204940986e-06,
+      "loss": 0.7923,
+      "step": 18430
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.0333288852378667e-06,
+      "loss": 0.8053,
+      "step": 18435
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.025907394956279e-06,
+      "loss": 0.781,
+      "step": 18440
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.0185122554055248e-06,
+      "loss": 0.7767,
+      "step": 18445
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.0111434723213697e-06,
+      "loss": 0.7832,
+      "step": 18450
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.0038010514191326e-06,
+      "loss": 0.7823,
+      "step": 18455
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 9.96484998393693e-07,
+      "loss": 0.7845,
+      "step": 18460
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 9.89195318919468e-07,
+      "loss": 0.7415,
+      "step": 18465
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 9.819320186504236e-07,
+      "loss": 0.7608,
+      "step": 18470
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 9.746951032200717e-07,
+      "loss": 0.8012,
+      "step": 18475
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 9.674845782414493e-07,
+      "loss": 0.7606,
+      "step": 18480
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 9.60300449307127e-07,
+      "loss": 0.7769,
+      "step": 18485
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 9.531427219892135e-07,
+      "loss": 0.737,
+      "step": 18490
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 9.460114018393233e-07,
+      "loss": 0.796,
+      "step": 18495
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 9.389064943886049e-07,
+      "loss": 0.8191,
+      "step": 18500
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 9.31828005147705e-07,
+      "loss": 0.7697,
+      "step": 18505
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 9.247759396067928e-07,
+      "loss": 0.7789,
+      "step": 18510
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 9.177503032355432e-07,
+      "loss": 0.8158,
+      "step": 18515
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 9.107511014831187e-07,
+      "loss": 0.7938,
+      "step": 18520
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 9.037783397781917e-07,
+      "loss": 0.7667,
+      "step": 18525
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.968320235289295e-07,
+      "loss": 0.7893,
+      "step": 18530
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.899121581229707e-07,
+      "loss": 0.751,
+      "step": 18535
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.830187489274526e-07,
+      "loss": 0.7476,
+      "step": 18540
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.761518012889956e-07,
+      "loss": 0.7721,
+      "step": 18545
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.693113205336838e-07,
+      "loss": 0.7991,
+      "step": 18550
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.624973119670803e-07,
+      "loss": 0.7752,
+      "step": 18555
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.557097808742158e-07,
+      "loss": 0.7368,
+      "step": 18560
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.489487325195927e-07,
+      "loss": 0.7753,
+      "step": 18565
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.422141721471531e-07,
+      "loss": 0.7936,
+      "step": 18570
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.355061049803035e-07,
+      "loss": 0.7777,
+      "step": 18575
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.288245362219176e-07,
+      "loss": 0.7745,
+      "step": 18580
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.221694710542975e-07,
+      "loss": 0.7945,
+      "step": 18585
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.155409146391939e-07,
+      "loss": 0.8042,
+      "step": 18590
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.089388721177975e-07,
+      "loss": 0.798,
+      "step": 18595
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.023633486107356e-07,
+      "loss": 0.8004,
+      "step": 18600
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 7.958143492180718e-07,
+      "loss": 0.7431,
+      "step": 18605
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 7.892918790192793e-07,
+      "loss": 0.7579,
+      "step": 18610
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 7.827959430732789e-07,
+      "loss": 0.7776,
+      "step": 18615
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 7.76326546418401e-07,
+      "loss": 0.7972,
+      "step": 18620
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 7.698836940723813e-07,
+      "loss": 0.7892,
+      "step": 18625
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 7.634673910323802e-07,
+      "loss": 0.7929,
+      "step": 18630
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 7.570776422749635e-07,
+      "loss": 0.7643,
+      "step": 18635
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 7.507144527561027e-07,
+      "loss": 0.7887,
+      "step": 18640
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 7.443778274111583e-07,
+      "loss": 0.7824,
+      "step": 18645
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 7.380677711549088e-07,
+      "loss": 0.8025,
+      "step": 18650
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 7.317842888815062e-07,
+      "loss": 0.8087,
+      "step": 18655
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 7.255273854644967e-07,
+      "loss": 0.7827,
+      "step": 18660
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 7.19297065756812e-07,
+      "loss": 0.7693,
+      "step": 18665
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 7.130933345907664e-07,
+      "loss": 0.7721,
+      "step": 18670
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 7.069161967780557e-07,
+      "loss": 0.785,
+      "step": 18675
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 7.007656571097425e-07,
+      "loss": 0.7774,
+      "step": 18680
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 6.946417203562599e-07,
+      "loss": 0.7646,
+      "step": 18685
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 6.885443912674149e-07,
+      "loss": 0.8133,
+      "step": 18690
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 6.824736745723658e-07,
+      "loss": 0.7637,
+      "step": 18695
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 6.764295749796372e-07,
+      "loss": 0.7617,
+      "step": 18700
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 6.704120971771165e-07,
+      "loss": 0.751,
+      "step": 18705
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 6.644212458320264e-07,
+      "loss": 0.7931,
+      "step": 18710
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 6.584570255909521e-07,
+      "loss": 0.7428,
+      "step": 18715
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 6.525194410798068e-07,
+      "loss": 0.7704,
+      "step": 18720
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 6.466084969038698e-07,
+      "loss": 0.7617,
+      "step": 18725
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 6.407241976477328e-07,
+      "loss": 0.7358,
+      "step": 18730
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 6.348665478753307e-07,
+      "loss": 0.7854,
+      "step": 18735
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 6.290355521299411e-07,
+      "loss": 0.7221,
+      "step": 18740
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 6.232312149341468e-07,
+      "loss": 0.7833,
+      "step": 18745
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 6.17453540789865e-07,
+      "loss": 0.8016,
+      "step": 18750
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 6.117025341783377e-07,
+      "loss": 0.8328,
+      "step": 18755
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 6.059781995601143e-07,
+      "loss": 0.7864,
+      "step": 18760
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 6.002805413750605e-07,
+      "loss": 0.7937,
+      "step": 18765
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.946095640423503e-07,
+      "loss": 0.7383,
+      "step": 18770
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.889652719604621e-07,
+      "loss": 0.7768,
+      "step": 18775
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.833476695071898e-07,
+      "loss": 0.7685,
+      "step": 18780
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.77756761039605e-07,
+      "loss": 0.7219,
+      "step": 18785
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.721925508940872e-07,
+      "loss": 0.7907,
+      "step": 18790
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.666550433863127e-07,
+      "loss": 0.7665,
+      "step": 18795
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.611442428112384e-07,
+      "loss": 0.7813,
+      "step": 18800
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.556601534431143e-07,
+      "loss": 0.7827,
+      "step": 18805
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.502027795354636e-07,
+      "loss": 0.7823,
+      "step": 18810
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.447721253210901e-07,
+      "loss": 0.7945,
+      "step": 18815
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.393681950120871e-07,
+      "loss": 0.7763,
+      "step": 18820
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.339909927998049e-07,
+      "loss": 0.7425,
+      "step": 18825
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.286405228548673e-07,
+      "loss": 0.7655,
+      "step": 18830
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.233167893271717e-07,
+      "loss": 0.772,
+      "step": 18835
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.180197963458649e-07,
+      "loss": 0.8331,
+      "step": 18840
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.127495480193633e-07,
+      "loss": 0.8148,
+      "step": 18845
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.075060484353449e-07,
+      "loss": 0.7948,
+      "step": 18850
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.022893016607221e-07,
+      "loss": 0.7215,
+      "step": 18855
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.970993117416728e-07,
+      "loss": 0.7564,
+      "step": 18860
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.919360827036207e-07,
+      "loss": 0.82,
+      "step": 18865
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.867996185512318e-07,
+      "loss": 0.7685,
+      "step": 18870
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.816899232684068e-07,
+      "loss": 0.7602,
+      "step": 18875
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.766070008182882e-07,
+      "loss": 0.787,
+      "step": 18880
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.715508551432645e-07,
+      "loss": 0.7564,
+      "step": 18885
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.6652149016493556e-07,
+      "loss": 0.7813,
+      "step": 18890
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.6151890978414317e-07,
+      "loss": 0.7677,
+      "step": 18895
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.565431178809481e-07,
+      "loss": 0.7712,
+      "step": 18900
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.5159411831464155e-07,
+      "loss": 0.7587,
+      "step": 18905
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.4667191492372577e-07,
+      "loss": 0.8114,
+      "step": 18910
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.4177651152592585e-07,
+      "loss": 0.7758,
+      "step": 18915
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.3690791191817784e-07,
+      "loss": 0.8108,
+      "step": 18920
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.32066119876629e-07,
+      "loss": 0.7519,
+      "step": 18925
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.272511391566297e-07,
+      "loss": 0.7286,
+      "step": 18930
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.224629734927415e-07,
+      "loss": 0.8005,
+      "step": 18935
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.177016265987254e-07,
+      "loss": 0.7615,
+      "step": 18940
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.129671021675457e-07,
+      "loss": 0.7964,
+      "step": 18945
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.082594038713466e-07,
+      "loss": 0.811,
+      "step": 18950
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.035785353614951e-07,
+      "loss": 0.7735,
+      "step": 18955
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.9892450026851484e-07,
+      "loss": 0.786,
+      "step": 18960
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.9429730220214837e-07,
+      "loss": 0.7725,
+      "step": 18965
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.8969694475129486e-07,
+      "loss": 0.7933,
+      "step": 18970
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.851234314840568e-07,
+      "loss": 0.7472,
+      "step": 18975
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.805767659477088e-07,
+      "loss": 0.7994,
+      "step": 18980
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.760569516686979e-07,
+      "loss": 0.7643,
+      "step": 18985
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.71563992152647e-07,
+      "loss": 0.7543,
+      "step": 18990
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.6709789088435917e-07,
+      "loss": 0.7686,
+      "step": 18995
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.6265865132779013e-07,
+      "loss": 0.7847,
+      "step": 19000
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.582462769260719e-07,
+      "loss": 0.8035,
+      "step": 19005
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.538607711015046e-07,
+      "loss": 0.7789,
+      "step": 19010
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.4950213725552953e-07,
+      "loss": 0.7974,
+      "step": 19015
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.451703787687643e-07,
+      "loss": 0.7708,
+      "step": 19020
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.4086549900097513e-07,
+      "loss": 0.7632,
+      "step": 19025
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.3658750129108124e-07,
+      "loss": 0.7487,
+      "step": 19030
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.3233638895714675e-07,
+      "loss": 0.7708,
+      "step": 19035
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.2811216529638864e-07,
+      "loss": 0.7763,
+      "step": 19040
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.2391483358517266e-07,
+      "loss": 0.7342,
+      "step": 19045
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.1974439707899815e-07,
+      "loss": 0.7449,
+      "step": 19050
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.156008590125053e-07,
+      "loss": 0.7769,
+      "step": 19055
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.114842225994757e-07,
+      "loss": 0.7673,
+      "step": 19060
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.073944910328241e-07,
+      "loss": 0.7837,
+      "step": 19065
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.033316674845987e-07,
+      "loss": 0.7506,
+      "step": 19070
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.9929575510597716e-07,
+      "loss": 0.7914,
+      "step": 19075
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.9528675702725877e-07,
+      "loss": 0.7835,
+      "step": 19080
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.9130467635788025e-07,
+      "loss": 0.7661,
+      "step": 19085
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.8734951618638815e-07,
+      "loss": 0.8005,
+      "step": 19090
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.834212795804508e-07,
+      "loss": 0.7793,
+      "step": 19095
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.795199695868661e-07,
+      "loss": 0.7356,
+      "step": 19100
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.7564558923153015e-07,
+      "loss": 0.7795,
+      "step": 19105
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.717981415194648e-07,
+      "loss": 0.8219,
+      "step": 19110
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.6797762943480183e-07,
+      "loss": 0.7847,
+      "step": 19115
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.641840559407715e-07,
+      "loss": 0.8043,
+      "step": 19120
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.604174239797219e-07,
+      "loss": 0.7835,
+      "step": 19125
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.566777364730954e-07,
+      "loss": 0.796,
+      "step": 19130
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.5296499632144474e-07,
+      "loss": 0.7594,
+      "step": 19135
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.492792064044169e-07,
+      "loss": 0.7487,
+      "step": 19140
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.4562036958075735e-07,
+      "loss": 0.8093,
+      "step": 19145
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.419884886882983e-07,
+      "loss": 0.7815,
+      "step": 19150
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.3838356654398593e-07,
+      "loss": 0.7523,
+      "step": 19155
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.348056059438258e-07,
+      "loss": 0.7883,
+      "step": 19160
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.312546096629414e-07,
+      "loss": 0.7536,
+      "step": 19165
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.2773058045552739e-07,
+      "loss": 0.7742,
+      "step": 19170
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.2423352105486513e-07,
+      "loss": 0.7869,
+      "step": 19175
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.207634341733111e-07,
+      "loss": 0.7614,
+      "step": 19180
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.1732032250231623e-07,
+      "loss": 0.7425,
+      "step": 19185
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.139041887123949e-07,
+      "loss": 0.8006,
+      "step": 19190
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.1051503545315206e-07,
+      "loss": 0.7717,
+      "step": 19195
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.071528653532445e-07,
+      "loss": 0.773,
+      "step": 19200
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.038176810204234e-07,
+      "loss": 0.7775,
+      "step": 19205
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.0050948504149567e-07,
+      "loss": 0.7469,
+      "step": 19210
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9722827998233937e-07,
+      "loss": 0.7662,
+      "step": 19215
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9397406838789987e-07,
+      "loss": 0.7808,
+      "step": 19220
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9074685278218204e-07,
+      "loss": 0.7585,
+      "step": 19225
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.8754663566825813e-07,
+      "loss": 0.7738,
+      "step": 19230
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.8437341952824814e-07,
+      "loss": 0.7704,
+      "step": 19235
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.8122720682335112e-07,
+      "loss": 0.7653,
+      "step": 19240
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.7810799999379844e-07,
+      "loss": 0.7823,
+      "step": 19245
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.750158014588926e-07,
+      "loss": 0.787,
+      "step": 19250
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.7195061361697237e-07,
+      "loss": 0.7719,
+      "step": 19255
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.689124388454477e-07,
+      "loss": 0.8113,
+      "step": 19260
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.6590127950075694e-07,
+      "loss": 0.7341,
+      "step": 19265
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.6291713791839412e-07,
+      "loss": 0.7677,
+      "step": 19270
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.5996001641290113e-07,
+      "loss": 0.7435,
+      "step": 19275
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.5702991727785218e-07,
+      "loss": 0.7935,
+      "step": 19280
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.541268427858694e-07,
+      "loss": 0.7934,
+      "step": 19285
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.512507951886188e-07,
+      "loss": 0.8004,
+      "step": 19290
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4840177671679498e-07,
+      "loss": 0.7637,
+      "step": 19295
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.455797895801325e-07,
+      "loss": 0.7778,
+      "step": 19300
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4278483596739832e-07,
+      "loss": 0.7625,
+      "step": 19305
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4001691804639948e-07,
+      "loss": 0.7476,
+      "step": 19310
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.372760379639598e-07,
+      "loss": 0.7623,
+      "step": 19315
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.3456219784594714e-07,
+      "loss": 0.8203,
+      "step": 19320
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.3187539979724215e-07,
+      "loss": 0.7638,
+      "step": 19325
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.2921564590176959e-07,
+      "loss": 0.7899,
+      "step": 19330
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.2658293822245923e-07,
+      "loss": 0.7662,
+      "step": 19335
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.239772788012733e-07,
+      "loss": 0.7951,
+      "step": 19340
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.2139866965919453e-07,
+      "loss": 0.7476,
+      "step": 19345
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.188471127962265e-07,
+      "loss": 0.7684,
+      "step": 19350
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.1632261019138944e-07,
+      "loss": 0.7986,
+      "step": 19355
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.1382516380271267e-07,
+      "loss": 0.7527,
+      "step": 19360
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.1135477556725004e-07,
+      "loss": 0.8059,
+      "step": 19365
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.0891144740107216e-07,
+      "loss": 0.8005,
+      "step": 19370
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.0649518119924316e-07,
+      "loss": 0.7977,
+      "step": 19375
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.0410597883585559e-07,
+      "loss": 0.7778,
+      "step": 19380
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.0174384216400711e-07,
+      "loss": 0.7474,
+      "step": 19385
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.940877301579663e-08,
+      "loss": 0.7827,
+      "step": 19390
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.7100773202336e-08,
+      "loss": 0.7821,
+      "step": 19395
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.481984451373436e-08,
+      "loss": 0.7649,
+      "step": 19400
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.256598871910992e-08,
+      "loss": 0.8113,
+      "step": 19405
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.033920756658597e-08,
+      "loss": 0.7769,
+      "step": 19410
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 8.813950278327154e-08,
+      "loss": 0.7823,
+      "step": 19415
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 8.596687607530018e-08,
+      "loss": 0.7863,
+      "step": 19420
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 8.382132912777173e-08,
+      "loss": 0.7389,
+      "step": 19425
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 8.170286360481449e-08,
+      "loss": 0.811,
+      "step": 19430
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.961148114952687e-08,
+      "loss": 0.8185,
+      "step": 19435
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.754718338401245e-08,
+      "loss": 0.8162,
+      "step": 19440
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.550997190936825e-08,
+      "loss": 0.7955,
+      "step": 19445
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.349984830568479e-08,
+      "loss": 0.8014,
+      "step": 19450
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.151681413203047e-08,
+      "loss": 0.765,
+      "step": 19455
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 6.956087092647888e-08,
+      "loss": 0.8012,
+      "step": 19460
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 6.763202020607761e-08,
+      "loss": 0.804,
+      "step": 19465
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 6.57302634668716e-08,
+      "loss": 0.7655,
+      "step": 19470
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 6.385560218388764e-08,
+      "loss": 0.8557,
+      "step": 19475
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 6.200803781113429e-08,
+      "loss": 0.7753,
+      "step": 19480
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 6.018757178160971e-08,
+      "loss": 0.7842,
+      "step": 19485
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.8394205507282244e-08,
+      "loss": 0.7845,
+      "step": 19490
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.662794037911755e-08,
+      "loss": 0.7765,
+      "step": 19495
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.488877776704759e-08,
+      "loss": 0.7806,
+      "step": 19500
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.317671901998999e-08,
+      "loss": 0.7739,
+      "step": 19505
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.149176546583644e-08,
+      "loss": 0.8072,
+      "step": 19510
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.9833918411464335e-08,
+      "loss": 0.7514,
+      "step": 19515
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.820317914271732e-08,
+      "loss": 0.7566,
+      "step": 19520
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.659954892441309e-08,
+      "loss": 0.7962,
+      "step": 19525
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.502302900035504e-08,
+      "loss": 0.7928,
+      "step": 19530
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.3473620593305056e-08,
+      "loss": 0.7978,
+      "step": 19535
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.1951324905006854e-08,
+      "loss": 0.7437,
+      "step": 19540
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.04561431161704e-08,
+      "loss": 0.7599,
+      "step": 19545
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.8988076386479716e-08,
+      "loss": 0.7471,
+      "step": 19550
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.754712585458508e-08,
+      "loss": 0.7888,
+      "step": 19555
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.613329263810305e-08,
+      "loss": 0.7919,
+      "step": 19560
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.474657783362034e-08,
+      "loss": 0.7765,
+      "step": 19565
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.338698251669769e-08,
+      "loss": 0.7867,
+      "step": 19570
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.205450774184271e-08,
+      "loss": 0.7527,
+      "step": 19575
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.074915454254867e-08,
+      "loss": 0.7509,
+      "step": 19580
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.9470923931259606e-08,
+      "loss": 0.7664,
+      "step": 19585
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.821981689938968e-08,
+      "loss": 0.7598,
+      "step": 19590
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.699583441731157e-08,
+      "loss": 0.7435,
+      "step": 19595
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.5798977434360323e-08,
+      "loss": 0.7622,
+      "step": 19600
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.462924687883727e-08,
+      "loss": 0.765,
+      "step": 19605
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.3486643657994463e-08,
+      "loss": 0.7941,
+      "step": 19610
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.237116865805022e-08,
+      "loss": 0.7614,
+      "step": 19615
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.1282822744185246e-08,
+      "loss": 0.8199,
+      "step": 19620
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.0221606760530972e-08,
+      "loss": 0.7989,
+      "step": 19625
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.9187521530181214e-08,
+      "loss": 0.8048,
+      "step": 19630
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8180567855184402e-08,
+      "loss": 0.7613,
+      "step": 19635
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.7200746516543572e-08,
+      "loss": 0.7745,
+      "step": 19640
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.6248058274228038e-08,
+      "loss": 0.805,
+      "step": 19645
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.5322503867146173e-08,
+      "loss": 0.7933,
+      "step": 19650
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4424084013172632e-08,
+      "loss": 0.8002,
+      "step": 19655
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.3552799409136673e-08,
+      "loss": 0.7632,
+      "step": 19660
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.2708650730814396e-08,
+      "loss": 0.8222,
+      "step": 19665
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.1891638632940404e-08,
+      "loss": 0.7644,
+      "step": 19670
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.1101763749192249e-08,
+      "loss": 0.7661,
+      "step": 19675
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.0339026692217644e-08,
+      "loss": 0.7801,
+      "step": 19680
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 9.603428053595597e-09,
+      "loss": 0.7749,
+      "step": 19685
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.894968403867497e-09,
+      "loss": 0.767,
+      "step": 19690
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.213648292529352e-09,
+      "loss": 0.794,
+      "step": 19695
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.559468248016232e-09,
+      "loss": 0.7715,
+      "step": 19700
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 6.9324287777178246e-09,
+      "loss": 0.7824,
+      "step": 19705
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 6.332530367978428e-09,
+      "loss": 0.7945,
+      "step": 19710
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 5.7597734840852945e-09,
+      "loss": 0.7838,
+      "step": 19715
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.214158570268634e-09,
+      "loss": 0.7973,
+      "step": 19720
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.695686049724923e-09,
+      "loss": 0.7503,
+      "step": 19725
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.204356324581937e-09,
+      "loss": 0.7595,
+      "step": 19730
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.740169775922064e-09,
+      "loss": 0.7805,
+      "step": 19735
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.303126763774533e-09,
+      "loss": 0.7971,
+      "step": 19740
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.893227627115413e-09,
+      "loss": 0.7499,
+      "step": 19745
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.5104726838676147e-09,
+      "loss": 0.7592,
+      "step": 19750
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.1548622308970033e-09,
+      "loss": 0.8055,
+      "step": 19755
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8263965440279416e-09,
+      "loss": 0.7747,
+      "step": 19760
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.5250758780122053e-09,
+      "loss": 0.7916,
+      "step": 19765
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.250900466567839e-09,
+      "loss": 0.7705,
+      "step": 19770
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.0038705223403e-09,
+      "loss": 0.7524,
+      "step": 19775
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.839862369374283e-10,
+      "loss": 0.7711,
+      "step": 19780
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 5.912477808944771e-10,
+      "loss": 0.7512,
+      "step": 19785
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.2565530370908263e-10,
+      "loss": 0.7821,
+      "step": 19790
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.8720893381795063e-10,
+      "loss": 0.7726,
+      "step": 19795
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.7590877859297026e-10,
+      "loss": 0.7508,
+      "step": 19800
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 9.175492436841458e-11,
+      "loss": 0.7505,
+      "step": 19805
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.474743641374011e-11,
+      "loss": 0.7611,
+      "step": 19810
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.886358941358359e-12,
+      "loss": 0.7713,
+      "step": 19815
+    },
+    {
+      "epoch": 3.0,
+      "eval_loss": 0.9637993574142456,
+      "eval_runtime": 236.092,
+      "eval_samples_per_second": 47.13,
+      "eval_steps_per_second": 5.892,
+      "step": 19818
+    },
+    {
+      "epoch": 3.0,
+      "step": 19818,
+      "total_flos": 1.6123402460420112e+19,
+      "train_loss": 0.9284211427668099,
+      "train_runtime": 113797.7954,
+      "train_samples_per_second": 11.145,
+      "train_steps_per_second": 0.174
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 19818,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 100,
+  "total_flos": 1.6123402460420112e+19,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}