diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,7219 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.030336661096518617,
+  "eval_steps": 500,
+  "global_step": 6000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 5e-06,
+      "loss": 1.7996,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1e-05,
+      "loss": 1.8749,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5e-05,
+      "loss": 1.7917,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2e-05,
+      "loss": 1.866,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5e-05,
+      "loss": 1.8179,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3e-05,
+      "loss": 1.7424,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.5e-05,
+      "loss": 1.776,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4e-05,
+      "loss": 1.7429,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.5e-05,
+      "loss": 1.7392,
+      "step": 45
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5e-05,
+      "loss": 1.7274,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.500000000000001e-05,
+      "loss": 1.7122,
+      "step": 55
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6e-05,
+      "loss": 1.7711,
+      "step": 60
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.500000000000001e-05,
+      "loss": 1.6894,
+      "step": 65
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7e-05,
+      "loss": 1.7153,
+      "step": 70
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 1.72,
+      "step": 75
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8e-05,
+      "loss": 1.6923,
+      "step": 80
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.5e-05,
+      "loss": 1.6293,
+      "step": 85
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9e-05,
+      "loss": 1.6991,
+      "step": 90
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.5e-05,
+      "loss": 1.6649,
+      "step": 95
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001,
+      "loss": 1.6809,
+      "step": 100
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999999984214647e-05,
+      "loss": 1.6536,
+      "step": 105
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999999936858588e-05,
+      "loss": 1.6798,
+      "step": 110
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999999857931825e-05,
+      "loss": 1.6354,
+      "step": 115
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999999747434355e-05,
+      "loss": 1.6531,
+      "step": 120
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999999605366182e-05,
+      "loss": 1.6346,
+      "step": 125
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999999431727304e-05,
+      "loss": 1.5923,
+      "step": 130
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999999226517724e-05,
+      "loss": 1.6328,
+      "step": 135
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999998989737443e-05,
+      "loss": 1.643,
+      "step": 140
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999998721386463e-05,
+      "loss": 1.6708,
+      "step": 145
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999998421464784e-05,
+      "loss": 1.6129,
+      "step": 150
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.99999808997241e-05,
+      "loss": 1.6173,
+      "step": 155
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999997726909342e-05,
+      "loss": 1.62,
+      "step": 160
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999997332275582e-05,
+      "loss": 1.6656,
+      "step": 165
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999996906071134e-05,
+      "loss": 1.5967,
+      "step": 170
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999996448295999e-05,
+      "loss": 1.6081,
+      "step": 175
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999995958950179e-05,
+      "loss": 1.6811,
+      "step": 180
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.99999543803368e-05,
+      "loss": 1.6281,
+      "step": 185
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999994885546504e-05,
+      "loss": 1.6343,
+      "step": 190
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999994301488653e-05,
+      "loss": 1.6014,
+      "step": 195
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999993685860133e-05,
+      "loss": 1.6031,
+      "step": 200
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999993038660947e-05,
+      "loss": 1.6057,
+      "step": 205
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999992359891099e-05,
+      "loss": 1.6461,
+      "step": 210
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999991649550593e-05,
+      "loss": 1.5893,
+      "step": 215
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999990907639434e-05,
+      "loss": 1.585,
+      "step": 220
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999990134157626e-05,
+      "loss": 1.6501,
+      "step": 225
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999989329105175e-05,
+      "loss": 1.585,
+      "step": 230
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999988492482087e-05,
+      "loss": 1.5825,
+      "step": 235
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999987624288363e-05,
+      "loss": 1.6001,
+      "step": 240
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999986724524012e-05,
+      "loss": 1.5628,
+      "step": 245
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999985793189038e-05,
+      "loss": 1.5727,
+      "step": 250
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999984830283449e-05,
+      "loss": 1.6333,
+      "step": 255
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.99998383580725e-05,
+      "loss": 1.6112,
+      "step": 260
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999982809760446e-05,
+      "loss": 1.5798,
+      "step": 265
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999981752143045e-05,
+      "loss": 1.5776,
+      "step": 270
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999980662955052e-05,
+      "loss": 1.5801,
+      "step": 275
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999979542196479e-05,
+      "loss": 1.6455,
+      "step": 280
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999978389867326e-05,
+      "loss": 1.5998,
+      "step": 285
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999977205967603e-05,
+      "loss": 1.6335,
+      "step": 290
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999975990497319e-05,
+      "loss": 1.5837,
+      "step": 295
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999974743456482e-05,
+      "loss": 1.6105,
+      "step": 300
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999973464845096e-05,
+      "loss": 1.5818,
+      "step": 305
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999972154663173e-05,
+      "loss": 1.602,
+      "step": 310
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.99997081291072e-05,
+      "loss": 1.5588,
+      "step": 315
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999969439587746e-05,
+      "loss": 1.596,
+      "step": 320
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999968034694258e-05,
+      "loss": 1.5618,
+      "step": 325
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999966598230266e-05,
+      "loss": 1.6504,
+      "step": 330
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999965130195779e-05,
+      "loss": 1.5724,
+      "step": 335
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999963630590805e-05,
+      "loss": 1.5604,
+      "step": 340
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999962099415356e-05,
+      "loss": 1.5777,
+      "step": 345
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.99996053666944e-05,
+      "loss": 1.6312,
+      "step": 350
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999958942353067e-05,
+      "loss": 1.5696,
+      "step": 355
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999957316466249e-05,
+      "loss": 1.5721,
+      "step": 360
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999955659008992e-05,
+      "loss": 1.6266,
+      "step": 365
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999953969981311e-05,
+      "loss": 1.5786,
+      "step": 370
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999952249383214e-05,
+      "loss": 1.5544,
+      "step": 375
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999950497214712e-05,
+      "loss": 1.6046,
+      "step": 380
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999948713475817e-05,
+      "loss": 1.6058,
+      "step": 385
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.99994689816654e-05,
+      "loss": 1.5973,
+      "step": 390
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999945051286892e-05,
+      "loss": 1.6181,
+      "step": 395
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999943172836885e-05,
+      "loss": 1.5921,
+      "step": 400
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.99994126281653e-05,
+      "loss": 1.625,
+      "step": 405
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999939321225842e-05,
+      "loss": 1.5971,
+      "step": 410
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999937348064829e-05,
+      "loss": 1.5907,
+      "step": 415
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999935343333508e-05,
+      "loss": 1.5694,
+      "step": 420
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999933307031887e-05,
+      "loss": 1.6243,
+      "step": 425
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999931239159983e-05,
+      "loss": 1.5286,
+      "step": 430
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999929139717806e-05,
+      "loss": 1.5546,
+      "step": 435
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999927008705372e-05,
+      "loss": 1.6108,
+      "step": 440
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.99992484612269e-05,
+      "loss": 1.4987,
+      "step": 445
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999922651969779e-05,
+      "loss": 1.5822,
+      "step": 450
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.99992042624665e-05,
+      "loss": 1.4994,
+      "step": 455
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999918168953317e-05,
+      "loss": 1.5675,
+      "step": 460
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999915880089796e-05,
+      "loss": 1.5846,
+      "step": 465
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999913559656097e-05,
+      "loss": 1.5625,
+      "step": 470
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999911207652242e-05,
+      "loss": 1.6224,
+      "step": 475
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999908824078239e-05,
+      "loss": 1.5623,
+      "step": 480
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999906408934107e-05,
+      "loss": 1.6009,
+      "step": 485
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999903962219859e-05,
+      "loss": 1.5425,
+      "step": 490
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999901483935512e-05,
+      "loss": 1.6444,
+      "step": 495
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.99989897408108e-05,
+      "loss": 1.5167,
+      "step": 500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999896432656581e-05,
+      "loss": 1.556,
+      "step": 505
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999893859662031e-05,
+      "loss": 1.5834,
+      "step": 510
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999891255097444e-05,
+      "loss": 1.5508,
+      "step": 515
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999888618962838e-05,
+      "loss": 1.5247,
+      "step": 520
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999885951258228e-05,
+      "loss": 1.5769,
+      "step": 525
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999883251983634e-05,
+      "loss": 1.5263,
+      "step": 530
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.99988052113907e-05,
+      "loss": 1.608,
+      "step": 535
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999877758724556e-05,
+      "loss": 1.5593,
+      "step": 540
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999874964740105e-05,
+      "loss": 1.5406,
+      "step": 545
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.99987213918574e-05,
+      "loss": 1.5401,
+      "step": 550
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999869282061476e-05,
+      "loss": 1.5681,
+      "step": 555
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999866393367331e-05,
+      "loss": 1.5728,
+      "step": 560
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999863473103324e-05,
+      "loss": 1.5974,
+      "step": 565
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999860521269473e-05,
+      "loss": 1.5599,
+      "step": 570
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999857537865795e-05,
+      "loss": 1.5956,
+      "step": 575
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999854522892314e-05,
+      "loss": 1.5796,
+      "step": 580
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999851476349042e-05,
+      "loss": 1.6238,
+      "step": 585
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999848398236005e-05,
+      "loss": 1.618,
+      "step": 590
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999845288553216e-05,
+      "loss": 1.4921,
+      "step": 595
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.9998421473007e-05,
+      "loss": 1.594,
+      "step": 600
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999838974478475e-05,
+      "loss": 1.561,
+      "step": 605
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.99983577008656e-05,
+      "loss": 1.609,
+      "step": 610
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999832534124976e-05,
+      "loss": 1.5937,
+      "step": 615
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999829266593744e-05,
+      "loss": 1.5282,
+      "step": 620
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999825967492884e-05,
+      "loss": 1.5998,
+      "step": 625
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999822636822416e-05,
+      "loss": 1.5592,
+      "step": 630
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999819274582363e-05,
+      "loss": 1.5475,
+      "step": 635
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999815880772745e-05,
+      "loss": 1.6313,
+      "step": 640
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999812455393582e-05,
+      "loss": 1.6535,
+      "step": 645
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.9998089984449e-05,
+      "loss": 1.5264,
+      "step": 650
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999805509926716e-05,
+      "loss": 1.5979,
+      "step": 655
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999801989839055e-05,
+      "loss": 1.5618,
+      "step": 660
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999798438181938e-05,
+      "loss": 1.6201,
+      "step": 665
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999794854955388e-05,
+      "loss": 1.5691,
+      "step": 670
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999791240159426e-05,
+      "loss": 1.6399,
+      "step": 675
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999787593794079e-05,
+      "loss": 1.5727,
+      "step": 680
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999783915859364e-05,
+      "loss": 1.5899,
+      "step": 685
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999780206355309e-05,
+      "loss": 1.5925,
+      "step": 690
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999776465281936e-05,
+      "loss": 1.5569,
+      "step": 695
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999772692639268e-05,
+      "loss": 1.5749,
+      "step": 700
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.99976888842733e-05,
+      "loss": 1.5409,
+      "step": 705
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999765052646145e-05,
+      "loss": 1.5865,
+      "step": 710
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999761185295738e-05,
+      "loss": 1.5857,
+      "step": 715
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999757286376131e-05,
+      "loss": 1.6028,
+      "step": 720
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999753355887351e-05,
+      "loss": 1.5525,
+      "step": 725
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999749393829425e-05,
+      "loss": 1.5298,
+      "step": 730
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999745400202373e-05,
+      "loss": 1.5837,
+      "step": 735
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999741375006223e-05,
+      "loss": 1.609,
+      "step": 740
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999737318241001e-05,
+      "loss": 1.5891,
+      "step": 745
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.99973322990673e-05,
+      "loss": 1.5623,
+      "step": 750
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.99972911000344e-05,
+      "loss": 1.5996,
+      "step": 755
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999724958531151e-05,
+      "loss": 1.5489,
+      "step": 760
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999720775489896e-05,
+      "loss": 1.6058,
+      "step": 765
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999716560879696e-05,
+      "loss": 1.5234,
+      "step": 770
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.99971231470058e-05,
+      "loss": 1.5619,
+      "step": 775
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999708036952576e-05,
+      "loss": 1.5611,
+      "step": 780
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999703727635708e-05,
+      "loss": 1.5747,
+      "step": 785
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999699386750005e-05,
+      "loss": 1.5776,
+      "step": 790
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999695014295494e-05,
+      "loss": 1.5548,
+      "step": 795
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999690610272204e-05,
+      "loss": 1.5526,
+      "step": 800
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999686174680161e-05,
+      "loss": 1.5164,
+      "step": 805
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999681707519393e-05,
+      "loss": 1.4783,
+      "step": 810
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999677208789928e-05,
+      "loss": 1.5452,
+      "step": 815
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999672678491796e-05,
+      "loss": 1.5135,
+      "step": 820
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999668116625025e-05,
+      "loss": 1.5256,
+      "step": 825
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999663523189644e-05,
+      "loss": 1.5538,
+      "step": 830
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999658898185681e-05,
+      "loss": 1.6071,
+      "step": 835
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999654241613166e-05,
+      "loss": 1.5695,
+      "step": 840
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999649553472128e-05,
+      "loss": 1.5224,
+      "step": 845
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999644833762599e-05,
+      "loss": 1.5752,
+      "step": 850
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999640082484602e-05,
+      "loss": 1.5129,
+      "step": 855
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999635299638174e-05,
+      "loss": 1.569,
+      "step": 860
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999630485223343e-05,
+      "loss": 1.5718,
+      "step": 865
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.99962563924014e-05,
+      "loss": 1.5711,
+      "step": 870
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999620761688595e-05,
+      "loss": 1.5173,
+      "step": 875
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999615852568738e-05,
+      "loss": 1.6075,
+      "step": 880
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999610911880599e-05,
+      "loss": 1.549,
+      "step": 885
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999605939624213e-05,
+      "loss": 1.5395,
+      "step": 890
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999600935799608e-05,
+      "loss": 1.5699,
+      "step": 895
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999595900406817e-05,
+      "loss": 1.6032,
+      "step": 900
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999590833445871e-05,
+      "loss": 1.5008,
+      "step": 905
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999585734916803e-05,
+      "loss": 1.5711,
+      "step": 910
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999580604819644e-05,
+      "loss": 1.5846,
+      "step": 915
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.99957544315443e-05,
+      "loss": 1.5641,
+      "step": 920
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999570249921189e-05,
+      "loss": 1.5393,
+      "step": 925
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999565025119955e-05,
+      "loss": 1.5406,
+      "step": 930
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999559768750761e-05,
+      "loss": 1.6012,
+      "step": 935
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999554480813642e-05,
+      "loss": 1.5826,
+      "step": 940
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.99954916130863e-05,
+      "loss": 1.5666,
+      "step": 945
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999543810235758e-05,
+      "loss": 1.5688,
+      "step": 950
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999538427595061e-05,
+      "loss": 1.5858,
+      "step": 955
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999533013386573e-05,
+      "loss": 1.539,
+      "step": 960
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999527567610328e-05,
+      "loss": 1.5403,
+      "step": 965
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999522090266357e-05,
+      "loss": 1.5762,
+      "step": 970
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999516581354701e-05,
+      "loss": 1.5501,
+      "step": 975
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.99951104087539e-05,
+      "loss": 1.5513,
+      "step": 980
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.99950546882846e-05,
+      "loss": 1.5601,
+      "step": 985
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999499865213948e-05,
+      "loss": 1.5821,
+      "step": 990
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999494230031887e-05,
+      "loss": 1.55,
+      "step": 995
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999488563282313e-05,
+      "loss": 1.5428,
+      "step": 1000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999482864965264e-05,
+      "loss": 1.6102,
+      "step": 1005
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999477135080772e-05,
+      "loss": 1.5822,
+      "step": 1010
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999471373628877e-05,
+      "loss": 1.5751,
+      "step": 1015
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999465580609615e-05,
+      "loss": 1.5276,
+      "step": 1020
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99945975602302e-05,
+      "loss": 1.5953,
+      "step": 1025
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999453899869129e-05,
+      "loss": 1.5923,
+      "step": 1030
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999448012147982e-05,
+      "loss": 1.4987,
+      "step": 1035
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999442092859614e-05,
+      "loss": 1.6472,
+      "step": 1040
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999436142004062e-05,
+      "loss": 1.5915,
+      "step": 1045
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999430159581365e-05,
+      "loss": 1.566,
+      "step": 1050
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999424145591561e-05,
+      "loss": 1.5354,
+      "step": 1055
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999418100034685e-05,
+      "loss": 1.5651,
+      "step": 1060
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999412022910779e-05,
+      "loss": 1.5543,
+      "step": 1065
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999405914219878e-05,
+      "loss": 1.5432,
+      "step": 1070
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999399773962024e-05,
+      "loss": 1.5405,
+      "step": 1075
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999393602137253e-05,
+      "loss": 1.4827,
+      "step": 1080
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999387398745605e-05,
+      "loss": 1.567,
+      "step": 1085
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999381163787119e-05,
+      "loss": 1.5851,
+      "step": 1090
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999374897261834e-05,
+      "loss": 1.5537,
+      "step": 1095
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999368599169791e-05,
+      "loss": 1.5684,
+      "step": 1100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999362269511028e-05,
+      "loss": 1.6324,
+      "step": 1105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999355908285586e-05,
+      "loss": 1.5908,
+      "step": 1110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999349515493504e-05,
+      "loss": 1.5818,
+      "step": 1115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999343091134825e-05,
+      "loss": 1.5015,
+      "step": 1120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999336635209587e-05,
+      "loss": 1.5314,
+      "step": 1125
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999330147717831e-05,
+      "loss": 1.6139,
+      "step": 1130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.9993236286596e-05,
+      "loss": 1.5631,
+      "step": 1135
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999317078034934e-05,
+      "loss": 1.5393,
+      "step": 1140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999310495843873e-05,
+      "loss": 1.5802,
+      "step": 1145
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999303882086459e-05,
+      "loss": 1.5265,
+      "step": 1150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999297236762736e-05,
+      "loss": 1.5805,
+      "step": 1155
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999290559872742e-05,
+      "loss": 1.4904,
+      "step": 1160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999283851416525e-05,
+      "loss": 1.522,
+      "step": 1165
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99927711139412e-05,
+      "loss": 1.5492,
+      "step": 1170
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999270339805577e-05,
+      "loss": 1.5396,
+      "step": 1175
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999263536650934e-05,
+      "loss": 1.5411,
+      "step": 1180
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999256701930235e-05,
+      "loss": 1.581,
+      "step": 1185
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999249835643522e-05,
+      "loss": 1.6091,
+      "step": 1190
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999242937790842e-05,
+      "loss": 1.5867,
+      "step": 1195
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999236008372235e-05,
+      "loss": 1.5334,
+      "step": 1200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999229047387746e-05,
+      "loss": 1.5239,
+      "step": 1205
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999222054837419e-05,
+      "loss": 1.5217,
+      "step": 1210
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999215030721298e-05,
+      "loss": 1.5833,
+      "step": 1215
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999207975039429e-05,
+      "loss": 1.5455,
+      "step": 1220
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999200887791853e-05,
+      "loss": 1.482,
+      "step": 1225
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999193768978617e-05,
+      "loss": 1.5395,
+      "step": 1230
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999186618599767e-05,
+      "loss": 1.5382,
+      "step": 1235
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999179436655346e-05,
+      "loss": 1.5346,
+      "step": 1240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999172223145399e-05,
+      "loss": 1.4847,
+      "step": 1245
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999164978069974e-05,
+      "loss": 1.5358,
+      "step": 1250
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999157701429116e-05,
+      "loss": 1.5671,
+      "step": 1255
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99915039322287e-05,
+      "loss": 1.5468,
+      "step": 1260
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999143053451282e-05,
+      "loss": 1.6142,
+      "step": 1265
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.9991356821144e-05,
+      "loss": 1.544,
+      "step": 1270
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999128279212268e-05,
+      "loss": 1.5805,
+      "step": 1275
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999120844744935e-05,
+      "loss": 1.6026,
+      "step": 1280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999113378712447e-05,
+      "loss": 1.5879,
+      "step": 1285
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999105881114852e-05,
+      "loss": 1.5617,
+      "step": 1290
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999098351952195e-05,
+      "loss": 1.5493,
+      "step": 1295
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999090791224527e-05,
+      "loss": 1.4812,
+      "step": 1300
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999083198931893e-05,
+      "loss": 1.5836,
+      "step": 1305
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999075575074341e-05,
+      "loss": 1.5515,
+      "step": 1310
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999067919651921e-05,
+      "loss": 1.5362,
+      "step": 1315
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999060232664681e-05,
+      "loss": 1.6015,
+      "step": 1320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999052514112668e-05,
+      "loss": 1.6118,
+      "step": 1325
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999044763995932e-05,
+      "loss": 1.5226,
+      "step": 1330
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999036982314521e-05,
+      "loss": 1.5084,
+      "step": 1335
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999029169068485e-05,
+      "loss": 1.5113,
+      "step": 1340
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999021324257873e-05,
+      "loss": 1.5519,
+      "step": 1345
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999013447882735e-05,
+      "loss": 1.5396,
+      "step": 1350
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999005539943119e-05,
+      "loss": 1.5679,
+      "step": 1355
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998997600439077e-05,
+      "loss": 1.5439,
+      "step": 1360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998989629370659e-05,
+      "loss": 1.52,
+      "step": 1365
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998981626737914e-05,
+      "loss": 1.558,
+      "step": 1370
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998973592540892e-05,
+      "loss": 1.4872,
+      "step": 1375
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998965526779647e-05,
+      "loss": 1.5766,
+      "step": 1380
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998957429454227e-05,
+      "loss": 1.5765,
+      "step": 1385
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998949300564684e-05,
+      "loss": 1.5404,
+      "step": 1390
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998941140111068e-05,
+      "loss": 1.55,
+      "step": 1395
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998932948093434e-05,
+      "loss": 1.5632,
+      "step": 1400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99892472451183e-05,
+      "loss": 1.5557,
+      "step": 1405
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998916469366311e-05,
+      "loss": 1.5449,
+      "step": 1410
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998908182656925e-05,
+      "loss": 1.5139,
+      "step": 1415
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998899864383728e-05,
+      "loss": 1.5281,
+      "step": 1420
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998891514546773e-05,
+      "loss": 1.5736,
+      "step": 1425
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998883133146111e-05,
+      "loss": 1.5173,
+      "step": 1430
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998874720181795e-05,
+      "loss": 1.5877,
+      "step": 1435
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998866275653877e-05,
+      "loss": 1.5667,
+      "step": 1440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99885779956241e-05,
+      "loss": 1.5054,
+      "step": 1445
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998849291907453e-05,
+      "loss": 1.556,
+      "step": 1450
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998840752689053e-05,
+      "loss": 1.5338,
+      "step": 1455
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998832181907267e-05,
+      "loss": 1.5424,
+      "step": 1460
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99882357956215e-05,
+      "loss": 1.5117,
+      "step": 1465
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998814945653754e-05,
+      "loss": 1.531,
+      "step": 1470
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998806280182135e-05,
+      "loss": 1.5302,
+      "step": 1475
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998797583147348e-05,
+      "loss": 1.5085,
+      "step": 1480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998788854549447e-05,
+      "loss": 1.5074,
+      "step": 1485
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998780094388487e-05,
+      "loss": 1.5279,
+      "step": 1490
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998771302664524e-05,
+      "loss": 1.5998,
+      "step": 1495
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998762479377613e-05,
+      "loss": 1.5471,
+      "step": 1500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99875362452781e-05,
+      "loss": 1.5421,
+      "step": 1505
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998744738115171e-05,
+      "loss": 1.5692,
+      "step": 1510
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99873582013975e-05,
+      "loss": 1.6098,
+      "step": 1515
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998726870601609e-05,
+      "loss": 1.5027,
+      "step": 1520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998717889500798e-05,
+      "loss": 1.5163,
+      "step": 1525
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998708876837377e-05,
+      "loss": 1.5064,
+      "step": 1530
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998699832611403e-05,
+      "loss": 1.5468,
+      "step": 1535
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998690756822931e-05,
+      "loss": 1.5748,
+      "step": 1540
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998681649472021e-05,
+      "loss": 1.5112,
+      "step": 1545
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99867251055873e-05,
+      "loss": 1.5103,
+      "step": 1550
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998663340083115e-05,
+      "loss": 1.5412,
+      "step": 1555
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998654138045231e-05,
+      "loss": 1.5342,
+      "step": 1560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998644904445143e-05,
+      "loss": 1.5451,
+      "step": 1565
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998635639282903e-05,
+      "loss": 1.5328,
+      "step": 1570
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998626342558571e-05,
+      "loss": 1.5288,
+      "step": 1575
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998617014272208e-05,
+      "loss": 1.5191,
+      "step": 1580
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998607654423871e-05,
+      "loss": 1.518,
+      "step": 1585
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99859826301362e-05,
+      "loss": 1.5382,
+      "step": 1590
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998588840041512e-05,
+      "loss": 1.5361,
+      "step": 1595
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99857938550761e-05,
+      "loss": 1.4716,
+      "step": 1600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998569899411972e-05,
+      "loss": 1.5346,
+      "step": 1605
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998560381754658e-05,
+      "loss": 1.5335,
+      "step": 1610
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998550832535727e-05,
+      "loss": 1.571,
+      "step": 1615
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99854125175524e-05,
+      "loss": 1.5688,
+      "step": 1620
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99853163941326e-05,
+      "loss": 1.5505,
+      "step": 1625
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998521995509845e-05,
+      "loss": 1.5353,
+      "step": 1630
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998512320045055e-05,
+      "loss": 1.5067,
+      "step": 1635
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998502613018952e-05,
+      "loss": 1.5349,
+      "step": 1640
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998492874431599e-05,
+      "loss": 1.567,
+      "step": 1645
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998483104283056e-05,
+      "loss": 1.511,
+      "step": 1650
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998473302573385e-05,
+      "loss": 1.5579,
+      "step": 1655
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998463469302647e-05,
+      "loss": 1.5335,
+      "step": 1660
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998453604470905e-05,
+      "loss": 1.5173,
+      "step": 1665
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998443708078222e-05,
+      "loss": 1.5424,
+      "step": 1670
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99843378012466e-05,
+      "loss": 1.5325,
+      "step": 1675
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998423820610282e-05,
+      "loss": 1.5752,
+      "step": 1680
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998413829535147e-05,
+      "loss": 1.5561,
+      "step": 1685
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998403806899324e-05,
+      "loss": 1.5617,
+      "step": 1690
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998393752702873e-05,
+      "loss": 1.5346,
+      "step": 1695
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998383666945859e-05,
+      "loss": 1.5467,
+      "step": 1700
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998373549628343e-05,
+      "loss": 1.4466,
+      "step": 1705
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998363400750392e-05,
+      "loss": 1.5568,
+      "step": 1710
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998353220312069e-05,
+      "loss": 1.5046,
+      "step": 1715
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998343008313437e-05,
+      "loss": 1.5542,
+      "step": 1720
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99833276475456e-05,
+      "loss": 1.5236,
+      "step": 1725
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998322489635507e-05,
+      "loss": 1.4776,
+      "step": 1730
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99831218295634e-05,
+      "loss": 1.614,
+      "step": 1735
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998301844717123e-05,
+      "loss": 1.5401,
+      "step": 1740
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998291474917923e-05,
+      "loss": 1.515,
+      "step": 1745
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998281073558804e-05,
+      "loss": 1.5409,
+      "step": 1750
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998270640639833e-05,
+      "loss": 1.567,
+      "step": 1755
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998260176161076e-05,
+      "loss": 1.5215,
+      "step": 1760
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998249680122599e-05,
+      "loss": 1.5309,
+      "step": 1765
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998239152524467e-05,
+      "loss": 1.4966,
+      "step": 1770
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998228593366747e-05,
+      "loss": 1.5245,
+      "step": 1775
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998218002649506e-05,
+      "loss": 1.4797,
+      "step": 1780
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998207380372812e-05,
+      "loss": 1.5137,
+      "step": 1785
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99819672653673e-05,
+      "loss": 1.5058,
+      "step": 1790
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998186041141329e-05,
+      "loss": 1.6036,
+      "step": 1795
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998175324186674e-05,
+      "loss": 1.5146,
+      "step": 1800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998164575672835e-05,
+      "loss": 1.5022,
+      "step": 1805
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998153795599879e-05,
+      "loss": 1.5059,
+      "step": 1810
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998142983967875e-05,
+      "loss": 1.5061,
+      "step": 1815
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99813214077689e-05,
+      "loss": 1.5005,
+      "step": 1820
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998121266026993e-05,
+      "loss": 1.5239,
+      "step": 1825
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998110359718253e-05,
+      "loss": 1.5135,
+      "step": 1830
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998099421850737e-05,
+      "loss": 1.547,
+      "step": 1835
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998088452424516e-05,
+      "loss": 1.5184,
+      "step": 1840
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99807745143966e-05,
+      "loss": 1.5437,
+      "step": 1845
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998066418896238e-05,
+      "loss": 1.5234,
+      "step": 1850
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998055354794316e-05,
+      "loss": 1.5954,
+      "step": 1855
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998044259133969e-05,
+      "loss": 1.5129,
+      "step": 1860
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998033131915266e-05,
+      "loss": 1.5716,
+      "step": 1865
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998021973138274e-05,
+      "loss": 1.4783,
+      "step": 1870
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998010782803066e-05,
+      "loss": 1.5515,
+      "step": 1875
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997999560909712e-05,
+      "loss": 1.543,
+      "step": 1880
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997988307458283e-05,
+      "loss": 1.4829,
+      "step": 1885
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99797702244885e-05,
+      "loss": 1.4991,
+      "step": 1890
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997965705881485e-05,
+      "loss": 1.5513,
+      "step": 1895
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99795435775626e-05,
+      "loss": 1.4615,
+      "step": 1900
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997942978073243e-05,
+      "loss": 1.5439,
+      "step": 1905
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99793156683251e-05,
+      "loss": 1.5779,
+      "step": 1910
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997920124034133e-05,
+      "loss": 1.5227,
+      "step": 1915
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99790864967818e-05,
+      "loss": 1.5125,
+      "step": 1920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997897143764727e-05,
+      "loss": 1.4916,
+      "step": 1925
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997885606293844e-05,
+      "loss": 1.544,
+      "step": 1930
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997874037265608e-05,
+      "loss": 1.5055,
+      "step": 1935
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99786243668009e-05,
+      "loss": 1.4588,
+      "step": 1940
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99785080453736e-05,
+      "loss": 1.5419,
+      "step": 1945
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997839140837497e-05,
+      "loss": 1.5481,
+      "step": 1950
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997827445580572e-05,
+      "loss": 1.6031,
+      "step": 1955
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997815718766658e-05,
+      "loss": 1.5483,
+      "step": 1960
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99780396039583e-05,
+      "loss": 1.5523,
+      "step": 1965
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997792170468162e-05,
+      "loss": 1.541,
+      "step": 1970
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997780348983728e-05,
+      "loss": 1.5834,
+      "step": 1975
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997768495942605e-05,
+      "loss": 1.5626,
+      "step": 1980
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997756611344864e-05,
+      "loss": 1.5052,
+      "step": 1985
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997744695190583e-05,
+      "loss": 1.5214,
+      "step": 1990
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997732747479837e-05,
+      "loss": 1.496,
+      "step": 1995
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997720768212701e-05,
+      "loss": 1.5063,
+      "step": 2000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99770875738925e-05,
+      "loss": 1.5716,
+      "step": 2005
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997696715009558e-05,
+      "loss": 1.491,
+      "step": 2010
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997684641073705e-05,
+      "loss": 1.5733,
+      "step": 2015
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997672535581768e-05,
+      "loss": 1.5072,
+      "step": 2020
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997660398533818e-05,
+      "loss": 1.5445,
+      "step": 2025
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997648229929935e-05,
+      "loss": 1.5033,
+      "step": 2030
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997636029770197e-05,
+      "loss": 1.5877,
+      "step": 2035
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997623798054679e-05,
+      "loss": 1.5401,
+      "step": 2040
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997611534783456e-05,
+      "loss": 1.518,
+      "step": 2045
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99759923995661e-05,
+      "loss": 1.5661,
+      "step": 2050
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997586913574217e-05,
+      "loss": 1.6024,
+      "step": 2055
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997574555636356e-05,
+      "loss": 1.5285,
+      "step": 2060
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997562166143102e-05,
+      "loss": 1.5251,
+      "step": 2065
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997549745094535e-05,
+      "loss": 1.513,
+      "step": 2070
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997537292490734e-05,
+      "loss": 1.5422,
+      "step": 2075
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997524808331775e-05,
+      "loss": 1.5595,
+      "step": 2080
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99751229261774e-05,
+      "loss": 1.6162,
+      "step": 2085
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997499745348708e-05,
+      "loss": 1.4848,
+      "step": 2090
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997487166524755e-05,
+      "loss": 1.5274,
+      "step": 2095
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997474556145963e-05,
+      "loss": 1.5433,
+      "step": 2100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997461914212411e-05,
+      "loss": 1.5322,
+      "step": 2105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997449240724179e-05,
+      "loss": 1.5405,
+      "step": 2110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997436535681348e-05,
+      "loss": 1.5629,
+      "step": 2115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997423799083995e-05,
+      "loss": 1.5265,
+      "step": 2120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997411030932205e-05,
+      "loss": 1.5918,
+      "step": 2125
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997398231226055e-05,
+      "loss": 1.513,
+      "step": 2130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997385399965627e-05,
+      "loss": 1.5206,
+      "step": 2135
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997372537151002e-05,
+      "loss": 1.5626,
+      "step": 2140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99735964278226e-05,
+      "loss": 1.5216,
+      "step": 2145
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997346716859486e-05,
+      "loss": 1.4917,
+      "step": 2150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997333759382757e-05,
+      "loss": 1.5318,
+      "step": 2155
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997320770352159e-05,
+      "loss": 1.5348,
+      "step": 2160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997307749767771e-05,
+      "loss": 1.5397,
+      "step": 2165
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997294697629676e-05,
+      "loss": 1.5519,
+      "step": 2170
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997281613937956e-05,
+      "loss": 1.5524,
+      "step": 2175
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997268498692696e-05,
+      "loss": 1.5372,
+      "step": 2180
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997255351893976e-05,
+      "loss": 1.6459,
+      "step": 2185
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997242173541882e-05,
+      "loss": 1.5363,
+      "step": 2190
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997228963636494e-05,
+      "loss": 1.5172,
+      "step": 2195
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997215722177896e-05,
+      "loss": 1.5398,
+      "step": 2200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997202449166172e-05,
+      "loss": 1.5388,
+      "step": 2205
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997189144601407e-05,
+      "loss": 1.5532,
+      "step": 2210
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997175808483686e-05,
+      "loss": 1.5264,
+      "step": 2215
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997162440813088e-05,
+      "loss": 1.5251,
+      "step": 2220
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997149041589704e-05,
+      "loss": 1.5257,
+      "step": 2225
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997135610813613e-05,
+      "loss": 1.4777,
+      "step": 2230
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997122148484902e-05,
+      "loss": 1.5145,
+      "step": 2235
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99710865460366e-05,
+      "loss": 1.5228,
+      "step": 2240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997095129169965e-05,
+      "loss": 1.4485,
+      "step": 2245
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997081572183907e-05,
+      "loss": 1.5493,
+      "step": 2250
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997067983645569e-05,
+      "loss": 1.5696,
+      "step": 2255
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99705436355504e-05,
+      "loss": 1.5178,
+      "step": 2260
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997040711912402e-05,
+      "loss": 1.502,
+      "step": 2265
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997027028717745e-05,
+      "loss": 1.5349,
+      "step": 2270
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997013313971154e-05,
+      "loss": 1.5049,
+      "step": 2275
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996999567672716e-05,
+      "loss": 1.5232,
+      "step": 2280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996985789822515e-05,
+      "loss": 1.5508,
+      "step": 2285
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996971980420642e-05,
+      "loss": 1.5511,
+      "step": 2290
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99695813946718e-05,
+      "loss": 1.5355,
+      "step": 2295
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996944266962222e-05,
+      "loss": 1.5373,
+      "step": 2300
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99693036290585e-05,
+      "loss": 1.563,
+      "step": 2305
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996916427298155e-05,
+      "loss": 1.5324,
+      "step": 2310
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996902460139226e-05,
+      "loss": 1.5672,
+      "step": 2315
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996888461429148e-05,
+      "loss": 1.5321,
+      "step": 2320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996874431168008e-05,
+      "loss": 1.5211,
+      "step": 2325
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.9968603693559e-05,
+      "loss": 1.4997,
+      "step": 2330
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99684627599291e-05,
+      "loss": 1.571,
+      "step": 2335
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996832151079127e-05,
+      "loss": 1.5507,
+      "step": 2340
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99681799461464e-05,
+      "loss": 1.5583,
+      "step": 2345
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99680380659954e-05,
+      "loss": 1.517,
+      "step": 2350
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996789587033912e-05,
+      "loss": 1.5097,
+      "step": 2355
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996775335917852e-05,
+      "loss": 1.5261,
+      "step": 2360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996761053251446e-05,
+      "loss": 1.5335,
+      "step": 2365
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996746739034786e-05,
+      "loss": 1.5407,
+      "step": 2370
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99673239326796e-05,
+      "loss": 1.5252,
+      "step": 2375
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996718015951061e-05,
+      "loss": 1.4937,
+      "step": 2380
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996703607084179e-05,
+      "loss": 1.5121,
+      "step": 2385
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996689166667406e-05,
+      "loss": 1.4867,
+      "step": 2390
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99667469470083e-05,
+      "loss": 1.5516,
+      "step": 2395
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996660191184546e-05,
+      "loss": 1.5321,
+      "step": 2400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996645656118644e-05,
+      "loss": 1.5535,
+      "step": 2405
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996631089503214e-05,
+      "loss": 1.5573,
+      "step": 2410
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996616491338352e-05,
+      "loss": 1.5682,
+      "step": 2415
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996601861624147e-05,
+      "loss": 1.5171,
+      "step": 2420
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996587200360692e-05,
+      "loss": 1.5273,
+      "step": 2425
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996572507548081e-05,
+      "loss": 1.4945,
+      "step": 2430
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996557783186406e-05,
+      "loss": 1.5208,
+      "step": 2435
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996543027275758e-05,
+      "loss": 1.529,
+      "step": 2440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996528239816233e-05,
+      "loss": 1.4699,
+      "step": 2445
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996513420807923e-05,
+      "loss": 1.5693,
+      "step": 2450
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996498570250922e-05,
+      "loss": 1.477,
+      "step": 2455
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996483688145324e-05,
+      "loss": 1.454,
+      "step": 2460
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996468774491222e-05,
+      "loss": 1.507,
+      "step": 2465
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99645382928871e-05,
+      "loss": 1.5086,
+      "step": 2470
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996438852537884e-05,
+      "loss": 1.4753,
+      "step": 2475
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996423844238836e-05,
+      "loss": 1.5156,
+      "step": 2480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996408804391663e-05,
+      "loss": 1.4741,
+      "step": 2485
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996393732996461e-05,
+      "loss": 1.5344,
+      "step": 2490
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996378630053322e-05,
+      "loss": 1.5078,
+      "step": 2495
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996363495562344e-05,
+      "loss": 1.5143,
+      "step": 2500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996348329523622e-05,
+      "loss": 1.5224,
+      "step": 2505
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996333131937248e-05,
+      "loss": 1.5095,
+      "step": 2510
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996317902803323e-05,
+      "loss": 1.451,
+      "step": 2515
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996302642121942e-05,
+      "loss": 1.5299,
+      "step": 2520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.9962873498932e-05,
+      "loss": 1.5151,
+      "step": 2525
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996272026117196e-05,
+      "loss": 1.5189,
+      "step": 2530
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996256670794022e-05,
+      "loss": 1.5397,
+      "step": 2535
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99624128392378e-05,
+      "loss": 1.5134,
+      "step": 2540
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996225865506564e-05,
+      "loss": 1.5024,
+      "step": 2545
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996210415542473e-05,
+      "loss": 1.5519,
+      "step": 2550
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996194934031604e-05,
+      "loss": 1.5404,
+      "step": 2555
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996179420974055e-05,
+      "loss": 1.4881,
+      "step": 2560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996163876369924e-05,
+      "loss": 1.5769,
+      "step": 2565
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996148300219308e-05,
+      "loss": 1.5814,
+      "step": 2570
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996132692522306e-05,
+      "loss": 1.4895,
+      "step": 2575
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996117053279017e-05,
+      "loss": 1.5461,
+      "step": 2580
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99610138248954e-05,
+      "loss": 1.5383,
+      "step": 2585
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996085680153974e-05,
+      "loss": 1.5121,
+      "step": 2590
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996069946272416e-05,
+      "loss": 1.5801,
+      "step": 2595
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996054180844968e-05,
+      "loss": 1.5253,
+      "step": 2600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996038383871729e-05,
+      "loss": 1.5258,
+      "step": 2605
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996022555352797e-05,
+      "loss": 1.5068,
+      "step": 2610
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996006695288273e-05,
+      "loss": 1.5297,
+      "step": 2615
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995990803678259e-05,
+      "loss": 1.5454,
+      "step": 2620
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995974880522853e-05,
+      "loss": 1.5164,
+      "step": 2625
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995958925822156e-05,
+      "loss": 1.5505,
+      "step": 2630
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995942939576268e-05,
+      "loss": 1.5168,
+      "step": 2635
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995926921785292e-05,
+      "loss": 1.5132,
+      "step": 2640
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995910872449327e-05,
+      "loss": 1.4964,
+      "step": 2645
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995894791568477e-05,
+      "loss": 1.5126,
+      "step": 2650
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99587867914284e-05,
+      "loss": 1.4778,
+      "step": 2655
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995862535172522e-05,
+      "loss": 1.5032,
+      "step": 2660
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995846359657622e-05,
+      "loss": 1.494,
+      "step": 2665
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995830152598241e-05,
+      "loss": 1.5445,
+      "step": 2670
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995813913994484e-05,
+      "loss": 1.5408,
+      "step": 2675
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995797643846453e-05,
+      "loss": 1.4772,
+      "step": 2680
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995781342154249e-05,
+      "loss": 1.5315,
+      "step": 2685
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995765008917977e-05,
+      "loss": 1.5888,
+      "step": 2690
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99574864413774e-05,
+      "loss": 1.4974,
+      "step": 2695
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99573224781364e-05,
+      "loss": 1.509,
+      "step": 2700
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995715819945783e-05,
+      "loss": 1.5193,
+      "step": 2705
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995699360534269e-05,
+      "loss": 1.528,
+      "step": 2710
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995682869579203e-05,
+      "loss": 1.5071,
+      "step": 2715
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995666347080692e-05,
+      "loss": 1.494,
+      "step": 2720
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995649793038837e-05,
+      "loss": 1.512,
+      "step": 2725
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995633207453745e-05,
+      "loss": 1.5239,
+      "step": 2730
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99561659032552e-05,
+      "loss": 1.5525,
+      "step": 2735
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995599941654266e-05,
+      "loss": 1.5474,
+      "step": 2740
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995583261440087e-05,
+      "loss": 1.557,
+      "step": 2745
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99556654968309e-05,
+      "loss": 1.5016,
+      "step": 2750
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995549806383383e-05,
+      "loss": 1.5156,
+      "step": 2755
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995533031541067e-05,
+      "loss": 1.4697,
+      "step": 2760
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995516225156251e-05,
+      "loss": 1.492,
+      "step": 2765
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995499387229041e-05,
+      "loss": 1.542,
+      "step": 2770
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995482517759543e-05,
+      "loss": 1.5223,
+      "step": 2775
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995465616747861e-05,
+      "loss": 1.5568,
+      "step": 2780
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995448684194104e-05,
+      "loss": 1.5111,
+      "step": 2785
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99543172009838e-05,
+      "loss": 1.5416,
+      "step": 2790
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995414724460793e-05,
+      "loss": 1.4793,
+      "step": 2795
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995397697281453e-05,
+      "loss": 1.4842,
+      "step": 2800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995380638560468e-05,
+      "loss": 1.4727,
+      "step": 2805
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995363548297943e-05,
+      "loss": 1.5372,
+      "step": 2810
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995346426493987e-05,
+      "loss": 1.486,
+      "step": 2815
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995329273148708e-05,
+      "loss": 1.5398,
+      "step": 2820
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995312088262216e-05,
+      "loss": 1.506,
+      "step": 2825
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995294871834617e-05,
+      "loss": 1.5192,
+      "step": 2830
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995277623866022e-05,
+      "loss": 1.5395,
+      "step": 2835
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995260344356539e-05,
+      "loss": 1.5055,
+      "step": 2840
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995243033306276e-05,
+      "loss": 1.5103,
+      "step": 2845
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995225690715344e-05,
+      "loss": 1.4474,
+      "step": 2850
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995208316583851e-05,
+      "loss": 1.5004,
+      "step": 2855
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995190910911907e-05,
+      "loss": 1.4629,
+      "step": 2860
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995173473699621e-05,
+      "loss": 1.5233,
+      "step": 2865
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995156004947107e-05,
+      "loss": 1.4981,
+      "step": 2870
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995138504654472e-05,
+      "loss": 1.4929,
+      "step": 2875
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995120972821828e-05,
+      "loss": 1.505,
+      "step": 2880
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995103409449282e-05,
+      "loss": 1.5656,
+      "step": 2885
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99508581453695e-05,
+      "loss": 1.5047,
+      "step": 2890
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995068188084942e-05,
+      "loss": 1.4868,
+      "step": 2895
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995050530093367e-05,
+      "loss": 1.5294,
+      "step": 2900
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995032840562338e-05,
+      "loss": 1.5434,
+      "step": 2905
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995015119491965e-05,
+      "loss": 1.533,
+      "step": 2910
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.994997366882361e-05,
+      "loss": 1.5523,
+      "step": 2915
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.994979582733642e-05,
+      "loss": 1.5377,
+      "step": 2920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.994961767045913e-05,
+      "loss": 1.4969,
+      "step": 2925
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.994943919819291e-05,
+      "loss": 1.5979,
+      "step": 2930
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.994926041053889e-05,
+      "loss": 1.5196,
+      "step": 2935
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.994908130749818e-05,
+      "loss": 1.5421,
+      "step": 2940
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.994890188907191e-05,
+      "loss": 1.5627,
+      "step": 2945
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.994872215526124e-05,
+      "loss": 1.5439,
+      "step": 2950
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.994854210606728e-05,
+      "loss": 1.5637,
+      "step": 2955
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.994836174149116e-05,
+      "loss": 1.465,
+      "step": 2960
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.994818106153402e-05,
+      "loss": 1.503,
+      "step": 2965
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994800006619705e-05,
+      "loss": 1.5463,
+      "step": 2970
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994781875548134e-05,
+      "loss": 1.5887,
+      "step": 2975
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994763712938804e-05,
+      "loss": 1.538,
+      "step": 2980
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994745518791832e-05,
+      "loss": 1.4573,
+      "step": 2985
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994727293107331e-05,
+      "loss": 1.5372,
+      "step": 2990
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994709035885417e-05,
+      "loss": 1.5234,
+      "step": 2995
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994690747126204e-05,
+      "loss": 1.4949,
+      "step": 3000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99467242682981e-05,
+      "loss": 1.5092,
+      "step": 3005
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994654074996348e-05,
+      "loss": 1.5308,
+      "step": 3010
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994635691625935e-05,
+      "loss": 1.5072,
+      "step": 3015
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994617276718686e-05,
+      "loss": 1.4754,
+      "step": 3020
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99459883027472e-05,
+      "loss": 1.4705,
+      "step": 3025
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994580352294152e-05,
+      "loss": 1.5202,
+      "step": 3030
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994561842777097e-05,
+      "loss": 1.5326,
+      "step": 3035
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994543301723674e-05,
+      "loss": 1.4853,
+      "step": 3040
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994524729134e-05,
+      "loss": 1.5118,
+      "step": 3045
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994506125008189e-05,
+      "loss": 1.4941,
+      "step": 3050
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994487489346364e-05,
+      "loss": 1.5047,
+      "step": 3055
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99446882214864e-05,
+      "loss": 1.5304,
+      "step": 3060
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994450123415133e-05,
+      "loss": 1.494,
+      "step": 3065
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994431393145965e-05,
+      "loss": 1.4773,
+      "step": 3070
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994412631341249e-05,
+      "loss": 1.5026,
+      "step": 3075
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994393838001108e-05,
+      "loss": 1.5637,
+      "step": 3080
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994375013125659e-05,
+      "loss": 1.5263,
+      "step": 3085
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994356156715022e-05,
+      "loss": 1.5137,
+      "step": 3090
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994337268769314e-05,
+      "loss": 1.5198,
+      "step": 3095
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994318349288655e-05,
+      "loss": 1.518,
+      "step": 3100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994299398273165e-05,
+      "loss": 1.5231,
+      "step": 3105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994280415722963e-05,
+      "loss": 1.5613,
+      "step": 3110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994261401638171e-05,
+      "loss": 1.4401,
+      "step": 3115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994242356018905e-05,
+      "loss": 1.4348,
+      "step": 3120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99422327886529e-05,
+      "loss": 1.5746,
+      "step": 3125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994204170177444e-05,
+      "loss": 1.5122,
+      "step": 3130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994185029955486e-05,
+      "loss": 1.518,
+      "step": 3135
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994165858199539e-05,
+      "loss": 1.4871,
+      "step": 3140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994146654909725e-05,
+      "loss": 1.4953,
+      "step": 3145
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994127420086161e-05,
+      "loss": 1.5215,
+      "step": 3150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994108153728973e-05,
+      "loss": 1.4941,
+      "step": 3155
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994088855838282e-05,
+      "loss": 1.5802,
+      "step": 3160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994069526414208e-05,
+      "loss": 1.5252,
+      "step": 3165
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994050165456874e-05,
+      "loss": 1.559,
+      "step": 3170
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994030772966403e-05,
+      "loss": 1.5166,
+      "step": 3175
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994011348942915e-05,
+      "loss": 1.4929,
+      "step": 3180
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993991893386534e-05,
+      "loss": 1.5256,
+      "step": 3185
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993972406297385e-05,
+      "loss": 1.49,
+      "step": 3190
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993952887675587e-05,
+      "loss": 1.555,
+      "step": 3195
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993933337521267e-05,
+      "loss": 1.4504,
+      "step": 3200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993913755834546e-05,
+      "loss": 1.5377,
+      "step": 3205
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993894142615549e-05,
+      "loss": 1.5529,
+      "step": 3210
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993874497864399e-05,
+      "loss": 1.4826,
+      "step": 3215
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99385482158122e-05,
+      "loss": 1.5276,
+      "step": 3220
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993835113766136e-05,
+      "loss": 1.4329,
+      "step": 3225
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993815374419273e-05,
+      "loss": 1.5641,
+      "step": 3230
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993795603540754e-05,
+      "loss": 1.5104,
+      "step": 3235
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993775801130706e-05,
+      "loss": 1.5004,
+      "step": 3240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99375596718925e-05,
+      "loss": 1.4776,
+      "step": 3245
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993736101716517e-05,
+      "loss": 1.4893,
+      "step": 3250
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993716204712626e-05,
+      "loss": 1.5128,
+      "step": 3255
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993696276177708e-05,
+      "loss": 1.5456,
+      "step": 3260
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993676316111887e-05,
+      "loss": 1.4825,
+      "step": 3265
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993656324515286e-05,
+      "loss": 1.5172,
+      "step": 3270
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993636301388036e-05,
+      "loss": 1.5193,
+      "step": 3275
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993616246730259e-05,
+      "loss": 1.5517,
+      "step": 3280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993596160542087e-05,
+      "loss": 1.5099,
+      "step": 3285
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993576042823642e-05,
+      "loss": 1.4761,
+      "step": 3290
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993555893575055e-05,
+      "loss": 1.5321,
+      "step": 3295
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993535712796447e-05,
+      "loss": 1.5451,
+      "step": 3300
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993515500487954e-05,
+      "loss": 1.5722,
+      "step": 3305
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993495256649695e-05,
+      "loss": 1.4602,
+      "step": 3310
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993474981281805e-05,
+      "loss": 1.5073,
+      "step": 3315
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993454674384407e-05,
+      "loss": 1.5392,
+      "step": 3320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993434335957632e-05,
+      "loss": 1.5624,
+      "step": 3325
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993413966001608e-05,
+      "loss": 1.5157,
+      "step": 3330
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993393564516462e-05,
+      "loss": 1.5243,
+      "step": 3335
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993373131502324e-05,
+      "loss": 1.5236,
+      "step": 3340
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993352666959324e-05,
+      "loss": 1.5148,
+      "step": 3345
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993332170887591e-05,
+      "loss": 1.5247,
+      "step": 3350
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99331164328725e-05,
+      "loss": 1.4501,
+      "step": 3355
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993291084158438e-05,
+      "loss": 1.5403,
+      "step": 3360
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99327049350128e-05,
+      "loss": 1.4707,
+      "step": 3365
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993249871315906e-05,
+      "loss": 1.4824,
+      "step": 3370
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99322921760245e-05,
+      "loss": 1.4943,
+      "step": 3375
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993208532361036e-05,
+      "loss": 1.5315,
+      "step": 3380
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993187815591801e-05,
+      "loss": 1.5344,
+      "step": 3385
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993167067294873e-05,
+      "loss": 1.5218,
+      "step": 3390
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993146287470383e-05,
+      "loss": 1.5238,
+      "step": 3395
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993125476118462e-05,
+      "loss": 1.5105,
+      "step": 3400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99310463323924e-05,
+      "loss": 1.5031,
+      "step": 3405
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993083758832853e-05,
+      "loss": 1.5065,
+      "step": 3410
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993062852899429e-05,
+      "loss": 1.4841,
+      "step": 3415
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.9930419154391e-05,
+      "loss": 1.5589,
+      "step": 3420
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993020946452002e-05,
+      "loss": 1.4913,
+      "step": 3425
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992999945938264e-05,
+      "loss": 1.516,
+      "step": 3430
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99297891389802e-05,
+      "loss": 1.5157,
+      "step": 3435
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992957850331399e-05,
+      "loss": 1.5128,
+      "step": 3440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99293675523854e-05,
+      "loss": 1.5151,
+      "step": 3445
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992915628619572e-05,
+      "loss": 1.4946,
+      "step": 3450
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99289447047463e-05,
+      "loss": 1.4794,
+      "step": 3455
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992873280803849e-05,
+      "loss": 1.4567,
+      "step": 3460
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992852059607358e-05,
+      "loss": 1.5259,
+      "step": 3465
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992830806885296e-05,
+      "loss": 1.5298,
+      "step": 3470
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992809522637795e-05,
+      "loss": 1.5511,
+      "step": 3475
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992788206864991e-05,
+      "loss": 1.5268,
+      "step": 3480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992766859567017e-05,
+      "loss": 1.4739,
+      "step": 3485
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992745480744008e-05,
+      "loss": 1.4906,
+      "step": 3490
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992724070396098e-05,
+      "loss": 1.5345,
+      "step": 3495
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992702628523422e-05,
+      "loss": 1.4536,
+      "step": 3500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99268115512612e-05,
+      "loss": 1.4975,
+      "step": 3505
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992659650204323e-05,
+      "loss": 1.5348,
+      "step": 3510
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992638113758168e-05,
+      "loss": 1.5162,
+      "step": 3515
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99261654578779e-05,
+      "loss": 1.5277,
+      "step": 3520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992594946293327e-05,
+      "loss": 1.5186,
+      "step": 3525
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992573315274914e-05,
+      "loss": 1.4776,
+      "step": 3530
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99255165273269e-05,
+      "loss": 1.502,
+      "step": 3535
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992529958666788e-05,
+      "loss": 1.5049,
+      "step": 3540
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992508233077348e-05,
+      "loss": 1.512,
+      "step": 3545
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992486475964506e-05,
+      "loss": 1.5675,
+      "step": 3550
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.9924646873284e-05,
+      "loss": 1.5572,
+      "step": 3555
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992442867169165e-05,
+      "loss": 1.5045,
+      "step": 3560
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992421015486943e-05,
+      "loss": 1.5,
+      "step": 3565
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992399132281869e-05,
+      "loss": 1.5228,
+      "step": 3570
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992377217554082e-05,
+      "loss": 1.516,
+      "step": 3575
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992355271303719e-05,
+      "loss": 1.4356,
+      "step": 3580
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992333293530922e-05,
+      "loss": 1.532,
+      "step": 3585
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992311284235827e-05,
+      "loss": 1.5011,
+      "step": 3590
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992289243418574e-05,
+      "loss": 1.4729,
+      "step": 3595
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992267171079301e-05,
+      "loss": 1.4935,
+      "step": 3600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99224506721815e-05,
+      "loss": 1.5192,
+      "step": 3605
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992222931835258e-05,
+      "loss": 1.5306,
+      "step": 3610
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992200764930763e-05,
+      "loss": 1.5288,
+      "step": 3615
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992178566504811e-05,
+      "loss": 1.5358,
+      "step": 3620
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992156336557536e-05,
+      "loss": 1.5034,
+      "step": 3625
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992134075089084e-05,
+      "loss": 1.4965,
+      "step": 3630
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99211178209959e-05,
+      "loss": 1.5002,
+      "step": 3635
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992089457589198e-05,
+      "loss": 1.4796,
+      "step": 3640
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99206710155805e-05,
+      "loss": 1.5193,
+      "step": 3645
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992044714006283e-05,
+      "loss": 1.5129,
+      "step": 3650
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992022294934042e-05,
+      "loss": 1.4832,
+      "step": 3655
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991999844341467e-05,
+      "loss": 1.4397,
+      "step": 3660
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.9919773622287e-05,
+      "loss": 1.4875,
+      "step": 3665
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991954848595883e-05,
+      "loss": 1.5253,
+      "step": 3670
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99193230344316e-05,
+      "loss": 1.4675,
+      "step": 3675
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991909726770671e-05,
+      "loss": 1.5533,
+      "step": 3680
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991887118578558e-05,
+      "loss": 1.5269,
+      "step": 3685
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991864478866966e-05,
+      "loss": 1.4913,
+      "step": 3690
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991841807636036e-05,
+      "loss": 1.4657,
+      "step": 3695
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991819104885912e-05,
+      "loss": 1.5898,
+      "step": 3700
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991796370616738e-05,
+      "loss": 1.471,
+      "step": 3705
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991773604828657e-05,
+      "loss": 1.5591,
+      "step": 3710
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991750807521811e-05,
+      "loss": 1.5143,
+      "step": 3715
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991727978696348e-05,
+      "loss": 1.5713,
+      "step": 3720
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991705118352408e-05,
+      "loss": 1.4789,
+      "step": 3725
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991682226490137e-05,
+      "loss": 1.4906,
+      "step": 3730
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991659303109681e-05,
+      "loss": 1.5641,
+      "step": 3735
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991636348211184e-05,
+      "loss": 1.5029,
+      "step": 3740
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99161336179479e-05,
+      "loss": 1.4743,
+      "step": 3745
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991590343860642e-05,
+      "loss": 1.5302,
+      "step": 3750
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99156729440889e-05,
+      "loss": 1.5319,
+      "step": 3755
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991544213439677e-05,
+      "loss": 1.5088,
+      "step": 3760
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991521100953148e-05,
+      "loss": 1.4649,
+      "step": 3765
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991497956949452e-05,
+      "loss": 1.5334,
+      "step": 3770
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99147478142873e-05,
+      "loss": 1.5279,
+      "step": 3775
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991451574391134e-05,
+      "loss": 1.516,
+      "step": 3780
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991428335836808e-05,
+      "loss": 1.4928,
+      "step": 3785
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991405065765898e-05,
+      "loss": 1.4405,
+      "step": 3790
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991381764178551e-05,
+      "loss": 1.5142,
+      "step": 3795
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991358431074915e-05,
+      "loss": 1.4963,
+      "step": 3800
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991335066455138e-05,
+      "loss": 1.5076,
+      "step": 3805
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991311670319368e-05,
+      "loss": 1.5718,
+      "step": 3810
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991288242667749e-05,
+      "loss": 1.4916,
+      "step": 3815
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991264783500431e-05,
+      "loss": 1.4561,
+      "step": 3820
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991241292817564e-05,
+      "loss": 1.5167,
+      "step": 3825
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991217770619294e-05,
+      "loss": 1.5802,
+      "step": 3830
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991194216905771e-05,
+      "loss": 1.542,
+      "step": 3835
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991170631677143e-05,
+      "loss": 1.4965,
+      "step": 3840
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991147014933557e-05,
+      "loss": 1.5709,
+      "step": 3845
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991123366675166e-05,
+      "loss": 1.5383,
+      "step": 3850
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991099686902117e-05,
+      "loss": 1.4743,
+      "step": 3855
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99107597561456e-05,
+      "loss": 1.5117,
+      "step": 3860
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991052232812644e-05,
+      "loss": 1.4882,
+      "step": 3865
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99102845849652e-05,
+      "loss": 1.5421,
+      "step": 3870
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991004652666338e-05,
+      "loss": 1.499,
+      "step": 3875
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990980815322247e-05,
+      "loss": 1.5019,
+      "step": 3880
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990956946464399e-05,
+      "loss": 1.5189,
+      "step": 3885
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990933046092944e-05,
+      "loss": 1.5,
+      "step": 3890
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990909114208033e-05,
+      "loss": 1.504,
+      "step": 3895
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990885150809817e-05,
+      "loss": 1.493,
+      "step": 3900
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990861155898448e-05,
+      "loss": 1.5292,
+      "step": 3905
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990837129474075e-05,
+      "loss": 1.4811,
+      "step": 3910
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990813071536855e-05,
+      "loss": 1.5442,
+      "step": 3915
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990788982086934e-05,
+      "loss": 1.5521,
+      "step": 3920
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990764861124467e-05,
+      "loss": 1.5455,
+      "step": 3925
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990740708649607e-05,
+      "loss": 1.5175,
+      "step": 3930
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990716524662506e-05,
+      "loss": 1.5023,
+      "step": 3935
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990692309163314e-05,
+      "loss": 1.5238,
+      "step": 3940
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990668062152189e-05,
+      "loss": 1.4671,
+      "step": 3945
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990643783629279e-05,
+      "loss": 1.4789,
+      "step": 3950
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99061947359474e-05,
+      "loss": 1.5161,
+      "step": 3955
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990595132048726e-05,
+      "loss": 1.4652,
+      "step": 3960
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99057075899139e-05,
+      "loss": 1.4732,
+      "step": 3965
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990546354422883e-05,
+      "loss": 1.5079,
+      "step": 3970
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990521918343362e-05,
+      "loss": 1.5139,
+      "step": 3975
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990497450752983e-05,
+      "loss": 1.4872,
+      "step": 3980
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990472951651898e-05,
+      "loss": 1.494,
+      "step": 3985
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990448421040262e-05,
+      "loss": 1.5164,
+      "step": 3990
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99042385891823e-05,
+      "loss": 1.4692,
+      "step": 3995
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990399265285956e-05,
+      "loss": 1.4927,
+      "step": 4000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990374640143599e-05,
+      "loss": 1.5416,
+      "step": 4005
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990349983491309e-05,
+      "loss": 1.5582,
+      "step": 4010
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990325295329246e-05,
+      "loss": 1.4653,
+      "step": 4015
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990300575657565e-05,
+      "loss": 1.5256,
+      "step": 4020
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990275824476421e-05,
+      "loss": 1.5464,
+      "step": 4025
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99025104178597e-05,
+      "loss": 1.5305,
+      "step": 4030
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990226227586371e-05,
+      "loss": 1.5468,
+      "step": 4035
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990201381877778e-05,
+      "loss": 1.486,
+      "step": 4040
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990176504660349e-05,
+      "loss": 1.5599,
+      "step": 4045
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990151595934242e-05,
+      "loss": 1.4714,
+      "step": 4050
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990126655699613e-05,
+      "loss": 1.4766,
+      "step": 4055
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990101683956619e-05,
+      "loss": 1.544,
+      "step": 4060
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990076680705418e-05,
+      "loss": 1.5168,
+      "step": 4065
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990051645946168e-05,
+      "loss": 1.5307,
+      "step": 4070
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990026579679029e-05,
+      "loss": 1.5073,
+      "step": 4075
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990001481904157e-05,
+      "loss": 1.4967,
+      "step": 4080
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.98997635262171e-05,
+      "loss": 1.5607,
+      "step": 4085
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989951191831849e-05,
+      "loss": 1.4891,
+      "step": 4090
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.98992599953473e-05,
+      "loss": 1.4783,
+      "step": 4095
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989900775730516e-05,
+      "loss": 1.4935,
+      "step": 4100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989875520419363e-05,
+      "loss": 1.4895,
+      "step": 4105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989850233601432e-05,
+      "loss": 1.4632,
+      "step": 4110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989824915276881e-05,
+      "loss": 1.4752,
+      "step": 4115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989799565445872e-05,
+      "loss": 1.5499,
+      "step": 4120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989774184108563e-05,
+      "loss": 1.428,
+      "step": 4125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989748771265114e-05,
+      "loss": 1.537,
+      "step": 4130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.98972332691569e-05,
+      "loss": 1.5157,
+      "step": 4135
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989697851060446e-05,
+      "loss": 1.4997,
+      "step": 4140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989672343699547e-05,
+      "loss": 1.4946,
+      "step": 4145
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989646804833152e-05,
+      "loss": 1.5161,
+      "step": 4150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989621234461422e-05,
+      "loss": 1.5054,
+      "step": 4155
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989595632584519e-05,
+      "loss": 1.5207,
+      "step": 4160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989569999202603e-05,
+      "loss": 1.4765,
+      "step": 4165
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989544334315841e-05,
+      "loss": 1.5545,
+      "step": 4170
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989518637924388e-05,
+      "loss": 1.5858,
+      "step": 4175
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989492910028412e-05,
+      "loss": 1.4902,
+      "step": 4180
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989467150628073e-05,
+      "loss": 1.4587,
+      "step": 4185
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989441359723535e-05,
+      "loss": 1.4603,
+      "step": 4190
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989415537314958e-05,
+      "loss": 1.507,
+      "step": 4195
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989389683402508e-05,
+      "loss": 1.4852,
+      "step": 4200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989363797986346e-05,
+      "loss": 1.5239,
+      "step": 4205
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989337881066639e-05,
+      "loss": 1.4857,
+      "step": 4210
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989311932643545e-05,
+      "loss": 1.5302,
+      "step": 4215
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989285952717234e-05,
+      "loss": 1.5101,
+      "step": 4220
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989259941287864e-05,
+      "loss": 1.5009,
+      "step": 4225
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989233898355603e-05,
+      "loss": 1.4918,
+      "step": 4230
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989207823920617e-05,
+      "loss": 1.5294,
+      "step": 4235
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989181717983066e-05,
+      "loss": 1.525,
+      "step": 4240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989155580543118e-05,
+      "loss": 1.4571,
+      "step": 4245
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989129411600938e-05,
+      "loss": 1.4515,
+      "step": 4250
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989103211156689e-05,
+      "loss": 1.4865,
+      "step": 4255
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989076979210539e-05,
+      "loss": 1.5006,
+      "step": 4260
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989050715762652e-05,
+      "loss": 1.529,
+      "step": 4265
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989024420813193e-05,
+      "loss": 1.5761,
+      "step": 4270
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988998094362332e-05,
+      "loss": 1.5111,
+      "step": 4275
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.98897173641023e-05,
+      "loss": 1.5537,
+      "step": 4280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988945346957059e-05,
+      "loss": 1.5933,
+      "step": 4285
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988918926002981e-05,
+      "loss": 1.581,
+      "step": 4290
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988892473548163e-05,
+      "loss": 1.4913,
+      "step": 4295
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988865989592777e-05,
+      "loss": 1.5118,
+      "step": 4300
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988839474136984e-05,
+      "loss": 1.4788,
+      "step": 4305
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988812927180956e-05,
+      "loss": 1.4853,
+      "step": 4310
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988786348724858e-05,
+      "loss": 1.456,
+      "step": 4315
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988759738768858e-05,
+      "loss": 1.5449,
+      "step": 4320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988733097313125e-05,
+      "loss": 1.4819,
+      "step": 4325
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988706424357827e-05,
+      "loss": 1.5171,
+      "step": 4330
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988679719903133e-05,
+      "loss": 1.5015,
+      "step": 4335
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988652983949209e-05,
+      "loss": 1.5576,
+      "step": 4340
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988626216496229e-05,
+      "loss": 1.5293,
+      "step": 4345
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988599417544356e-05,
+      "loss": 1.4875,
+      "step": 4350
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988572587093761e-05,
+      "loss": 1.497,
+      "step": 4355
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988545725144616e-05,
+      "loss": 1.4931,
+      "step": 4360
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988518831697089e-05,
+      "loss": 1.4963,
+      "step": 4365
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988491906751351e-05,
+      "loss": 1.5466,
+      "step": 4370
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988464950307568e-05,
+      "loss": 1.5052,
+      "step": 4375
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988437962365915e-05,
+      "loss": 1.505,
+      "step": 4380
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988410942926561e-05,
+      "loss": 1.5476,
+      "step": 4385
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988383891989673e-05,
+      "loss": 1.5194,
+      "step": 4390
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988356809555427e-05,
+      "loss": 1.479,
+      "step": 4395
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.98832969562399e-05,
+      "loss": 1.5269,
+      "step": 4400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988302550195537e-05,
+      "loss": 1.5161,
+      "step": 4405
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988275373270237e-05,
+      "loss": 1.5332,
+      "step": 4410
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988248164848262e-05,
+      "loss": 1.4943,
+      "step": 4415
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988220924929784e-05,
+      "loss": 1.5402,
+      "step": 4420
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988193653514973e-05,
+      "loss": 1.5296,
+      "step": 4425
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988166350604004e-05,
+      "loss": 1.5358,
+      "step": 4430
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.98813901619705e-05,
+      "loss": 1.4855,
+      "step": 4435
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988111650294278e-05,
+      "loss": 1.4435,
+      "step": 4440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988084252895868e-05,
+      "loss": 1.4716,
+      "step": 4445
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.98805682400199e-05,
+      "loss": 1.4794,
+      "step": 4450
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988029363612815e-05,
+      "loss": 1.5067,
+      "step": 4455
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.98800187172852e-05,
+      "loss": 1.5073,
+      "step": 4460
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987974348349276e-05,
+      "loss": 1.5499,
+      "step": 4465
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987946793475257e-05,
+      "loss": 1.4601,
+      "step": 4470
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987919207106639e-05,
+      "loss": 1.5021,
+      "step": 4475
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987891589243594e-05,
+      "loss": 1.5213,
+      "step": 4480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987863939886298e-05,
+      "loss": 1.5026,
+      "step": 4485
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987836259034925e-05,
+      "loss": 1.5198,
+      "step": 4490
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.98780854668965e-05,
+      "loss": 1.477,
+      "step": 4495
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987780802850646e-05,
+      "loss": 1.4909,
+      "step": 4500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.98775302751809e-05,
+      "loss": 1.5267,
+      "step": 4505
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987725220692157e-05,
+      "loss": 1.4857,
+      "step": 4510
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987697382373025e-05,
+      "loss": 1.5284,
+      "step": 4515
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987669512560865e-05,
+      "loss": 1.4978,
+      "step": 4520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987641611255857e-05,
+      "loss": 1.5073,
+      "step": 4525
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987613678458174e-05,
+      "loss": 1.4503,
+      "step": 4530
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987585714167995e-05,
+      "loss": 1.4891,
+      "step": 4535
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987557718385497e-05,
+      "loss": 1.4578,
+      "step": 4540
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987529691110853e-05,
+      "loss": 1.4722,
+      "step": 4545
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987501632344244e-05,
+      "loss": 1.4837,
+      "step": 4550
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987473542085845e-05,
+      "loss": 1.5406,
+      "step": 4555
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987445420335833e-05,
+      "loss": 1.5159,
+      "step": 4560
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987417267094388e-05,
+      "loss": 1.5105,
+      "step": 4565
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987389082361684e-05,
+      "loss": 1.5373,
+      "step": 4570
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987360866137903e-05,
+      "loss": 1.5489,
+      "step": 4575
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987332618423221e-05,
+      "loss": 1.5048,
+      "step": 4580
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987304339217815e-05,
+      "loss": 1.4216,
+      "step": 4585
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987276028521867e-05,
+      "loss": 1.4706,
+      "step": 4590
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987247686335555e-05,
+      "loss": 1.4979,
+      "step": 4595
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987219312659055e-05,
+      "loss": 1.507,
+      "step": 4600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987190907492549e-05,
+      "loss": 1.5173,
+      "step": 4605
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987162470836215e-05,
+      "loss": 1.4988,
+      "step": 4610
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987134002690233e-05,
+      "loss": 1.4874,
+      "step": 4615
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987105503054783e-05,
+      "loss": 1.4652,
+      "step": 4620
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987076971930045e-05,
+      "loss": 1.4972,
+      "step": 4625
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987048409316199e-05,
+      "loss": 1.52,
+      "step": 4630
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987019815213425e-05,
+      "loss": 1.4863,
+      "step": 4635
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.986991189621902e-05,
+      "loss": 1.4899,
+      "step": 4640
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.986962532541814e-05,
+      "loss": 1.457,
+      "step": 4645
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.986933843973341e-05,
+      "loss": 1.5102,
+      "step": 4650
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.986905123916664e-05,
+      "loss": 1.4892,
+      "step": 4655
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.986876372371963e-05,
+      "loss": 1.5298,
+      "step": 4660
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.98684758933942e-05,
+      "loss": 1.5681,
+      "step": 4665
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.986818774819218e-05,
+      "loss": 1.4885,
+      "step": 4670
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.986789928811538e-05,
+      "loss": 1.4887,
+      "step": 4675
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.986761051316563e-05,
+      "loss": 1.5291,
+      "step": 4680
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.986732142334471e-05,
+      "loss": 1.5125,
+      "step": 4685
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.986703201865453e-05,
+      "loss": 1.5467,
+      "step": 4690
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.986674229909683e-05,
+      "loss": 1.5131,
+      "step": 4695
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.98664522646735e-05,
+      "loss": 1.474,
+      "step": 4700
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.986616191538635e-05,
+      "loss": 1.5111,
+      "step": 4705
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.98658712512372e-05,
+      "loss": 1.5524,
+      "step": 4710
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.986558027222793e-05,
+      "loss": 1.5388,
+      "step": 4715
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.986528897836032e-05,
+      "loss": 1.528,
+      "step": 4720
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.986499736963624e-05,
+      "loss": 1.4662,
+      "step": 4725
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.98647054460575e-05,
+      "loss": 1.5473,
+      "step": 4730
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.986441320762601e-05,
+      "loss": 1.5541,
+      "step": 4735
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.986412065434355e-05,
+      "loss": 1.5309,
+      "step": 4740
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.9863827786212e-05,
+      "loss": 1.4741,
+      "step": 4745
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.986353460323321e-05,
+      "loss": 1.442,
+      "step": 4750
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.986324110540901e-05,
+      "loss": 1.4524,
+      "step": 4755
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.986294729274127e-05,
+      "loss": 1.4878,
+      "step": 4760
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.986265316523184e-05,
+      "loss": 1.5478,
+      "step": 4765
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.986235872288256e-05,
+      "loss": 1.4916,
+      "step": 4770
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.986206396569533e-05,
+      "loss": 1.4551,
+      "step": 4775
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.986176889367198e-05,
+      "loss": 1.5169,
+      "step": 4780
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.986147350681439e-05,
+      "loss": 1.5001,
+      "step": 4785
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.986117780512441e-05,
+      "loss": 1.4939,
+      "step": 4790
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.986088178860391e-05,
+      "loss": 1.4576,
+      "step": 4795
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.986058545725476e-05,
+      "loss": 1.5747,
+      "step": 4800
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.986028881107882e-05,
+      "loss": 1.4927,
+      "step": 4805
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.985999185007802e-05,
+      "loss": 1.5061,
+      "step": 4810
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.985969457425414e-05,
+      "loss": 1.4841,
+      "step": 4815
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.985939698360916e-05,
+      "loss": 1.5298,
+      "step": 4820
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.985909907814487e-05,
+      "loss": 1.4187,
+      "step": 4825
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.98588008578632e-05,
+      "loss": 1.4835,
+      "step": 4830
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.985850232276603e-05,
+      "loss": 1.5325,
+      "step": 4835
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.985820347285521e-05,
+      "loss": 1.5165,
+      "step": 4840
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.985790430813269e-05,
+      "loss": 1.5617,
+      "step": 4845
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.98576048286003e-05,
+      "loss": 1.4625,
+      "step": 4850
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.985730503425997e-05,
+      "loss": 1.4876,
+      "step": 4855
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.985700492511356e-05,
+      "loss": 1.515,
+      "step": 4860
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.985670450116297e-05,
+      "loss": 1.5216,
+      "step": 4865
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.985640376241014e-05,
+      "loss": 1.4706,
+      "step": 4870
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.985610270885692e-05,
+      "loss": 1.4992,
+      "step": 4875
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.985580134050522e-05,
+      "loss": 1.4712,
+      "step": 4880
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.985549965735698e-05,
+      "loss": 1.5588,
+      "step": 4885
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.985519765941405e-05,
+      "loss": 1.4993,
+      "step": 4890
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.985489534667837e-05,
+      "loss": 1.5217,
+      "step": 4895
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.985459271915185e-05,
+      "loss": 1.5097,
+      "step": 4900
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.985428977683638e-05,
+      "loss": 1.478,
+      "step": 4905
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.985398651973389e-05,
+      "loss": 1.4965,
+      "step": 4910
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.98536829478463e-05,
+      "loss": 1.527,
+      "step": 4915
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.98533790611755e-05,
+      "loss": 1.5296,
+      "step": 4920
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.985307485972344e-05,
+      "loss": 1.5321,
+      "step": 4925
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.9852770343492e-05,
+      "loss": 1.4849,
+      "step": 4930
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.985246551248317e-05,
+      "loss": 1.4683,
+      "step": 4935
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.98521603666988e-05,
+      "loss": 1.5257,
+      "step": 4940
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.985185490614086e-05,
+      "loss": 1.5138,
+      "step": 4945
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.985154913081127e-05,
+      "loss": 1.5458,
+      "step": 4950
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.985124304071198e-05,
+      "loss": 1.5363,
+      "step": 4955
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.985093663584488e-05,
+      "loss": 1.5001,
+      "step": 4960
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.985062991621193e-05,
+      "loss": 1.4756,
+      "step": 4965
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.985032288181506e-05,
+      "loss": 1.5139,
+      "step": 4970
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.985001553265623e-05,
+      "loss": 1.5024,
+      "step": 4975
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984970786873735e-05,
+      "loss": 1.5113,
+      "step": 4980
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984939989006038e-05,
+      "loss": 1.4691,
+      "step": 4985
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984909159662727e-05,
+      "loss": 1.5188,
+      "step": 4990
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984878298843994e-05,
+      "loss": 1.4678,
+      "step": 4995
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984847406550037e-05,
+      "loss": 1.4888,
+      "step": 5000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984816482781048e-05,
+      "loss": 1.4616,
+      "step": 5005
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984785527537226e-05,
+      "loss": 1.5665,
+      "step": 5010
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984754540818763e-05,
+      "loss": 1.4752,
+      "step": 5015
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984723522625856e-05,
+      "loss": 1.4979,
+      "step": 5020
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.9846924729587e-05,
+      "loss": 1.5286,
+      "step": 5025
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984661391817494e-05,
+      "loss": 1.4881,
+      "step": 5030
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984630279202432e-05,
+      "loss": 1.4904,
+      "step": 5035
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.98459913511371e-05,
+      "loss": 1.4129,
+      "step": 5040
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984567959551526e-05,
+      "loss": 1.555,
+      "step": 5045
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984536752516074e-05,
+      "loss": 1.495,
+      "step": 5050
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984505514007555e-05,
+      "loss": 1.4924,
+      "step": 5055
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984474244026163e-05,
+      "loss": 1.5308,
+      "step": 5060
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984442942572099e-05,
+      "loss": 1.4795,
+      "step": 5065
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984411609645557e-05,
+      "loss": 1.5113,
+      "step": 5070
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984380245246736e-05,
+      "loss": 1.4893,
+      "step": 5075
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984348849375836e-05,
+      "loss": 1.5274,
+      "step": 5080
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984317422033052e-05,
+      "loss": 1.5588,
+      "step": 5085
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984285963218585e-05,
+      "loss": 1.5325,
+      "step": 5090
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984254472932633e-05,
+      "loss": 1.5095,
+      "step": 5095
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984222951175393e-05,
+      "loss": 1.5129,
+      "step": 5100
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984191397947067e-05,
+      "loss": 1.4927,
+      "step": 5105
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.98415981324785e-05,
+      "loss": 1.5628,
+      "step": 5110
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984128197077947e-05,
+      "loss": 1.4961,
+      "step": 5115
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984096549437555e-05,
+      "loss": 1.481,
+      "step": 5120
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984064870326872e-05,
+      "loss": 1.5379,
+      "step": 5125
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984033159746102e-05,
+      "loss": 1.5377,
+      "step": 5130
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984001417695442e-05,
+      "loss": 1.4382,
+      "step": 5135
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983969644175091e-05,
+      "loss": 1.5358,
+      "step": 5140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983937839185255e-05,
+      "loss": 1.5165,
+      "step": 5145
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983906002726131e-05,
+      "loss": 1.5192,
+      "step": 5150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.98387413479792e-05,
+      "loss": 1.5313,
+      "step": 5155
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983842235400824e-05,
+      "loss": 1.4434,
+      "step": 5160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983810304535047e-05,
+      "loss": 1.5003,
+      "step": 5165
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983778342200785e-05,
+      "loss": 1.5131,
+      "step": 5170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983746348398244e-05,
+      "loss": 1.487,
+      "step": 5175
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983714323127625e-05,
+      "loss": 1.4931,
+      "step": 5180
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.98368226638913e-05,
+      "loss": 1.4297,
+      "step": 5185
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.98365017818296e-05,
+      "loss": 1.5028,
+      "step": 5190
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983618058509321e-05,
+      "loss": 1.473,
+      "step": 5195
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983585907368413e-05,
+      "loss": 1.5346,
+      "step": 5200
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.98355372476044e-05,
+      "loss": 1.5364,
+      "step": 5205
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983521510685606e-05,
+      "loss": 1.5042,
+      "step": 5210
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983489265144112e-05,
+      "loss": 1.5007,
+      "step": 5215
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983456988136164e-05,
+      "loss": 1.5043,
+      "step": 5220
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983424679661966e-05,
+      "loss": 1.4609,
+      "step": 5225
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.98339233972172e-05,
+      "loss": 1.5366,
+      "step": 5230
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983359968315631e-05,
+      "loss": 1.5686,
+      "step": 5235
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983327565443906e-05,
+      "loss": 1.4633,
+      "step": 5240
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983295131106744e-05,
+      "loss": 1.4903,
+      "step": 5245
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983262665304353e-05,
+      "loss": 1.4585,
+      "step": 5250
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.98323016803694e-05,
+      "loss": 1.5265,
+      "step": 5255
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983197639304706e-05,
+      "loss": 1.482,
+      "step": 5260
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983165079107859e-05,
+      "loss": 1.5468,
+      "step": 5265
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983132487446605e-05,
+      "loss": 1.4755,
+      "step": 5270
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983099864321149e-05,
+      "loss": 1.51,
+      "step": 5275
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983067209731695e-05,
+      "loss": 1.5016,
+      "step": 5280
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983034523678454e-05,
+      "loss": 1.4736,
+      "step": 5285
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983001806161627e-05,
+      "loss": 1.4539,
+      "step": 5290
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.982969057181423e-05,
+      "loss": 1.5463,
+      "step": 5295
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.98293627673805e-05,
+      "loss": 1.4902,
+      "step": 5300
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.982903464831714e-05,
+      "loss": 1.5275,
+      "step": 5305
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.982870621462621e-05,
+      "loss": 1.5566,
+      "step": 5310
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.98283774663098e-05,
+      "loss": 1.4929,
+      "step": 5315
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.982804840336998e-05,
+      "loss": 1.4763,
+      "step": 5320
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.982771902580883e-05,
+      "loss": 1.5196,
+      "step": 5325
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.982738933362842e-05,
+      "loss": 1.5326,
+      "step": 5330
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.982705932683085e-05,
+      "loss": 1.5162,
+      "step": 5335
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.982672900541817e-05,
+      "loss": 1.5219,
+      "step": 5340
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.98263983693925e-05,
+      "loss": 1.5499,
+      "step": 5345
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.982606741875592e-05,
+      "loss": 1.5681,
+      "step": 5350
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.98257361535105e-05,
+      "loss": 1.5491,
+      "step": 5355
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.982540457365836e-05,
+      "loss": 1.5337,
+      "step": 5360
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.982507267920158e-05,
+      "loss": 1.5557,
+      "step": 5365
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.982474047014226e-05,
+      "loss": 1.5216,
+      "step": 5370
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.982440794648249e-05,
+      "loss": 1.506,
+      "step": 5375
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.982407510822437e-05,
+      "loss": 1.4518,
+      "step": 5380
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.982374195537001e-05,
+      "loss": 1.4318,
+      "step": 5385
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.98234084879215e-05,
+      "loss": 1.4863,
+      "step": 5390
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.982307470588098e-05,
+      "loss": 1.5002,
+      "step": 5395
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.98227406092505e-05,
+      "loss": 1.4875,
+      "step": 5400
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.982240619803221e-05,
+      "loss": 1.5092,
+      "step": 5405
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.982207147222822e-05,
+      "loss": 1.5387,
+      "step": 5410
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.982173643184063e-05,
+      "loss": 1.5063,
+      "step": 5415
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.982140107687156e-05,
+      "loss": 1.5373,
+      "step": 5420
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.982106540732312e-05,
+      "loss": 1.4991,
+      "step": 5425
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.982072942319745e-05,
+      "loss": 1.5448,
+      "step": 5430
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.982039312449666e-05,
+      "loss": 1.5134,
+      "step": 5435
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.982005651122288e-05,
+      "loss": 1.5076,
+      "step": 5440
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.981971958337824e-05,
+      "loss": 1.5199,
+      "step": 5445
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.981938234096482e-05,
+      "loss": 1.4118,
+      "step": 5450
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.981904478398482e-05,
+      "loss": 1.5121,
+      "step": 5455
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.981870691244031e-05,
+      "loss": 1.4717,
+      "step": 5460
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.981836872633348e-05,
+      "loss": 1.4902,
+      "step": 5465
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.981803022566641e-05,
+      "loss": 1.5429,
+      "step": 5470
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.981769141044127e-05,
+      "loss": 1.5351,
+      "step": 5475
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.98173522806602e-05,
+      "loss": 1.4532,
+      "step": 5480
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.981701283632532e-05,
+      "loss": 1.4673,
+      "step": 5485
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.98166730774388e-05,
+      "loss": 1.4805,
+      "step": 5490
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.981633300400277e-05,
+      "loss": 1.5073,
+      "step": 5495
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.981599261601939e-05,
+      "loss": 1.4875,
+      "step": 5500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.981565191349078e-05,
+      "loss": 1.4679,
+      "step": 5505
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.981531089641912e-05,
+      "loss": 1.5476,
+      "step": 5510
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.981496956480655e-05,
+      "loss": 1.4786,
+      "step": 5515
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.981462791865524e-05,
+      "loss": 1.5164,
+      "step": 5520
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.981428595796731e-05,
+      "loss": 1.5029,
+      "step": 5525
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.981394368274497e-05,
+      "loss": 1.4879,
+      "step": 5530
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.981360109299034e-05,
+      "loss": 1.4773,
+      "step": 5535
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.98132581887056e-05,
+      "loss": 1.4871,
+      "step": 5540
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.981291496989294e-05,
+      "loss": 1.5306,
+      "step": 5545
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.981257143655447e-05,
+      "loss": 1.4711,
+      "step": 5550
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.98122275886924e-05,
+      "loss": 1.5054,
+      "step": 5555
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.98118834263089e-05,
+      "loss": 1.4881,
+      "step": 5560
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.981153894940614e-05,
+      "loss": 1.5215,
+      "step": 5565
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.981119415798628e-05,
+      "loss": 1.5511,
+      "step": 5570
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.981084905205149e-05,
+      "loss": 1.5274,
+      "step": 5575
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.981050363160399e-05,
+      "loss": 1.5059,
+      "step": 5580
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.981015789664593e-05,
+      "loss": 1.5123,
+      "step": 5585
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.98098118471795e-05,
+      "loss": 1.4691,
+      "step": 5590
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.980946548320689e-05,
+      "loss": 1.4681,
+      "step": 5595
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.980911880473027e-05,
+      "loss": 1.401,
+      "step": 5600
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.980877181175186e-05,
+      "loss": 1.5356,
+      "step": 5605
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.980842450427382e-05,
+      "loss": 1.5111,
+      "step": 5610
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.980807688229836e-05,
+      "loss": 1.51,
+      "step": 5615
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.980772894582766e-05,
+      "loss": 1.4631,
+      "step": 5620
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.980738069486394e-05,
+      "loss": 1.4893,
+      "step": 5625
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.980703212940938e-05,
+      "loss": 1.4753,
+      "step": 5630
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.980668324946619e-05,
+      "loss": 1.4929,
+      "step": 5635
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.980633405503656e-05,
+      "loss": 1.4866,
+      "step": 5640
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.980598454612271e-05,
+      "loss": 1.5493,
+      "step": 5645
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.980563472272684e-05,
+      "loss": 1.4831,
+      "step": 5650
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.980528458485117e-05,
+      "loss": 1.4877,
+      "step": 5655
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.98049341324979e-05,
+      "loss": 1.4614,
+      "step": 5660
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.980458336566923e-05,
+      "loss": 1.4293,
+      "step": 5665
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.98042322843674e-05,
+      "loss": 1.5266,
+      "step": 5670
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.98038808885946e-05,
+      "loss": 1.4845,
+      "step": 5675
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.980352917835308e-05,
+      "loss": 1.4924,
+      "step": 5680
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.980317715364505e-05,
+      "loss": 1.4654,
+      "step": 5685
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.980282481447272e-05,
+      "loss": 1.3969,
+      "step": 5690
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.980247216083832e-05,
+      "loss": 1.5392,
+      "step": 5695
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.980211919274407e-05,
+      "loss": 1.4587,
+      "step": 5700
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.980176591019222e-05,
+      "loss": 1.5106,
+      "step": 5705
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.980141231318498e-05,
+      "loss": 1.5285,
+      "step": 5710
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.980105840172461e-05,
+      "loss": 1.5293,
+      "step": 5715
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.980070417581331e-05,
+      "loss": 1.4841,
+      "step": 5720
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.980034963545333e-05,
+      "loss": 1.4763,
+      "step": 5725
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.97999947806469e-05,
+      "loss": 1.4851,
+      "step": 5730
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.97996396113963e-05,
+      "loss": 1.5057,
+      "step": 5735
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.979928412770373e-05,
+      "loss": 1.5023,
+      "step": 5740
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.979892832957145e-05,
+      "loss": 1.4776,
+      "step": 5745
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.979857221700171e-05,
+      "loss": 1.5119,
+      "step": 5750
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.979821578999675e-05,
+      "loss": 1.5539,
+      "step": 5755
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.979785904855882e-05,
+      "loss": 1.4454,
+      "step": 5760
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.979750199269018e-05,
+      "loss": 1.4928,
+      "step": 5765
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.979714462239309e-05,
+      "loss": 1.4733,
+      "step": 5770
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.979678693766979e-05,
+      "loss": 1.469,
+      "step": 5775
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.979642893852255e-05,
+      "loss": 1.5397,
+      "step": 5780
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.979607062495362e-05,
+      "loss": 1.5132,
+      "step": 5785
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.979571199696527e-05,
+      "loss": 1.5642,
+      "step": 5790
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.979535305455977e-05,
+      "loss": 1.5103,
+      "step": 5795
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.979499379773936e-05,
+      "loss": 1.5058,
+      "step": 5800
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.979463422650634e-05,
+      "loss": 1.5033,
+      "step": 5805
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.979427434086299e-05,
+      "loss": 1.5078,
+      "step": 5810
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.979391414081153e-05,
+      "loss": 1.4872,
+      "step": 5815
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.97935536263543e-05,
+      "loss": 1.5118,
+      "step": 5820
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.97931927974935e-05,
+      "loss": 1.4501,
+      "step": 5825
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.979283165423148e-05,
+      "loss": 1.4908,
+      "step": 5830
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.979247019657049e-05,
+      "loss": 1.4134,
+      "step": 5835
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.979210842451281e-05,
+      "loss": 1.4692,
+      "step": 5840
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.979174633806072e-05,
+      "loss": 1.4497,
+      "step": 5845
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.979138393721653e-05,
+      "loss": 1.4799,
+      "step": 5850
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.979102122198248e-05,
+      "loss": 1.4355,
+      "step": 5855
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.97906581923609e-05,
+      "loss": 1.4799,
+      "step": 5860
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.97902948483541e-05,
+      "loss": 1.4916,
+      "step": 5865
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.978993118996432e-05,
+      "loss": 1.5522,
+      "step": 5870
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.97895672171939e-05,
+      "loss": 1.4886,
+      "step": 5875
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.978920293004513e-05,
+      "loss": 1.5207,
+      "step": 5880
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.978883832852028e-05,
+      "loss": 1.5016,
+      "step": 5885
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.97884734126217e-05,
+      "loss": 1.5223,
+      "step": 5890
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.978810818235164e-05,
+      "loss": 1.5014,
+      "step": 5895
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.978774263771247e-05,
+      "loss": 1.437,
+      "step": 5900
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.978737677870645e-05,
+      "loss": 1.4876,
+      "step": 5905
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.978701060533589e-05,
+      "loss": 1.5068,
+      "step": 5910
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.978664411760312e-05,
+      "loss": 1.5635,
+      "step": 5915
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.978627731551046e-05,
+      "loss": 1.5147,
+      "step": 5920
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.97859101990602e-05,
+      "loss": 1.5375,
+      "step": 5925
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.978554276825469e-05,
+      "loss": 1.5361,
+      "step": 5930
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.978517502309622e-05,
+      "loss": 1.4982,
+      "step": 5935
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.978480696358714e-05,
+      "loss": 1.4991,
+      "step": 5940
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.978443858972974e-05,
+      "loss": 1.5021,
+      "step": 5945
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.978406990152637e-05,
+      "loss": 1.4616,
+      "step": 5950
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.978370089897938e-05,
+      "loss": 1.5133,
+      "step": 5955
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.978333158209105e-05,
+      "loss": 1.4992,
+      "step": 5960
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.978296195086375e-05,
+      "loss": 1.4619,
+      "step": 5965
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.978259200529978e-05,
+      "loss": 1.4875,
+      "step": 5970
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.97822217454015e-05,
+      "loss": 1.4738,
+      "step": 5975
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.978185117117125e-05,
+      "loss": 1.5233,
+      "step": 5980
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.978148028261136e-05,
+      "loss": 1.4409,
+      "step": 5985
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.978110907972417e-05,
+      "loss": 1.5485,
+      "step": 5990
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.978073756251204e-05,
+      "loss": 1.5185,
+      "step": 5995
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.978036573097729e-05,
+      "loss": 1.515,
+      "step": 6000
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 197780,
+  "num_train_epochs": 1,
+  "save_steps": 1000,
+  "total_flos": 1.6825629945102336e+19,
+  "trial_name": null,
+  "trial_params": null
+}