diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,15381 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.015184011236168315,
+  "eval_steps": 500,
+  "global_step": 12800,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999999999957118,
+      "loss": 3.6491,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999999999827435,
+      "loss": 3.5313,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999999999610949,
+      "loss": 3.4388,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999999993076608,
+      "loss": 3.1787,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999999989175698,
+      "loss": 3.0979,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999999984406761,
+      "loss": 3.183,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000199999999787698,
+      "loss": 3.24,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999999972264808,
+      "loss": 3.2554,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999999964891794,
+      "loss": 3.1969,
+      "step": 45
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999999956650752,
+      "loss": 3.013,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999999947541687,
+      "loss": 2.9786,
+      "step": 55
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999999937564593,
+      "loss": 3.3584,
+      "step": 60
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999999926719473,
+      "loss": 3.3326,
+      "step": 65
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999999915006324,
+      "loss": 3.1324,
+      "step": 70
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999999990242515,
+      "loss": 3.3203,
+      "step": 75
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999999888975953,
+      "loss": 3.05,
+      "step": 80
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999999874658727,
+      "loss": 2.9144,
+      "step": 85
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999999859473478,
+      "loss": 3.1871,
+      "step": 90
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999999843420198,
+      "loss": 3.1406,
+      "step": 95
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999999826498895,
+      "loss": 2.7888,
+      "step": 100
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999999808709563,
+      "loss": 2.8994,
+      "step": 105
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999999790052206,
+      "loss": 3.0518,
+      "step": 110
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999999770526822,
+      "loss": 3.0918,
+      "step": 115
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999999750133416,
+      "loss": 3.0768,
+      "step": 120
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999999728871978,
+      "loss": 2.9068,
+      "step": 125
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999999706742517,
+      "loss": 3.1022,
+      "step": 130
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999999968374503,
+      "loss": 3.0395,
+      "step": 135
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999999659879515,
+      "loss": 2.9104,
+      "step": 140
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999999635145975,
+      "loss": 3.0261,
+      "step": 145
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999999609544408,
+      "loss": 2.9319,
+      "step": 150
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999999583074815,
+      "loss": 3.0491,
+      "step": 155
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999999555737197,
+      "loss": 3.0118,
+      "step": 160
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999999952753155,
+      "loss": 2.9509,
+      "step": 165
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999999949845788,
+      "loss": 3.0506,
+      "step": 170
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999999946851618,
+      "loss": 2.9915,
+      "step": 175
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999999437706457,
+      "loss": 3.0174,
+      "step": 180
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999999406028707,
+      "loss": 3.0289,
+      "step": 185
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999999937348293,
+      "loss": 2.9562,
+      "step": 190
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999999340069126,
+      "loss": 3.1903,
+      "step": 195
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999999305787301,
+      "loss": 2.9751,
+      "step": 200
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999999270637443,
+      "loss": 2.8893,
+      "step": 205
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999999923461956,
+      "loss": 2.8216,
+      "step": 210
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999999197733653,
+      "loss": 3.0853,
+      "step": 215
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999999159979722,
+      "loss": 3.1333,
+      "step": 220
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999999121357763,
+      "loss": 2.9536,
+      "step": 225
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999999081867775,
+      "loss": 2.937,
+      "step": 230
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999999041509764,
+      "loss": 2.8704,
+      "step": 235
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999999000283722,
+      "loss": 3.06,
+      "step": 240
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999999895818966,
+      "loss": 2.8869,
+      "step": 245
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999998915227568,
+      "loss": 2.8376,
+      "step": 250
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999999887139745,
+      "loss": 2.8362,
+      "step": 255
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999998826699308,
+      "loss": 2.9185,
+      "step": 260
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999998781133137,
+      "loss": 2.8776,
+      "step": 265
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999998734698943,
+      "loss": 2.9575,
+      "step": 270
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999998687396722,
+      "loss": 3.0368,
+      "step": 275
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999999863922647,
+      "loss": 2.8667,
+      "step": 280
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000199999985901882,
+      "loss": 2.8613,
+      "step": 285
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000199999985402819,
+      "loss": 2.9864,
+      "step": 290
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999998489507573,
+      "loss": 2.6927,
+      "step": 295
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999999843786522,
+      "loss": 2.7765,
+      "step": 300
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999999838535484,
+      "loss": 2.8588,
+      "step": 305
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999998331976434,
+      "loss": 2.7963,
+      "step": 310
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999998277730005,
+      "loss": 2.7799,
+      "step": 315
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999999822261555,
+      "loss": 2.9825,
+      "step": 320
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999998166633063,
+      "loss": 2.9661,
+      "step": 325
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999998109782553,
+      "loss": 2.821,
+      "step": 330
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999999805206402,
+      "loss": 2.8383,
+      "step": 335
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999997993477457,
+      "loss": 2.9595,
+      "step": 340
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999999793402287,
+      "loss": 2.8529,
+      "step": 345
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999997873700257,
+      "loss": 2.7759,
+      "step": 350
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999997812509616,
+      "loss": 2.8958,
+      "step": 355
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999999775045095,
+      "loss": 2.9969,
+      "step": 360
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999997687524256,
+      "loss": 2.9635,
+      "step": 365
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999999762372954,
+      "loss": 3.0617,
+      "step": 370
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999997559066795,
+      "loss": 2.7889,
+      "step": 375
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999997493536027,
+      "loss": 2.9124,
+      "step": 380
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999997427137229,
+      "loss": 2.866,
+      "step": 385
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999997359870407,
+      "loss": 2.8843,
+      "step": 390
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999997291735556,
+      "loss": 2.8506,
+      "step": 395
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999997222732682,
+      "loss": 3.0383,
+      "step": 400
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999997152861783,
+      "loss": 2.8744,
+      "step": 405
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999997082122858,
+      "loss": 2.8794,
+      "step": 410
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999997010515904,
+      "loss": 2.8376,
+      "step": 415
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999996938040927,
+      "loss": 2.8557,
+      "step": 420
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999996864697922,
+      "loss": 2.6925,
+      "step": 425
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999999679048689,
+      "loss": 2.5931,
+      "step": 430
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999996715407833,
+      "loss": 3.012,
+      "step": 435
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999999663946075,
+      "loss": 2.6536,
+      "step": 440
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999996562645645,
+      "loss": 2.8957,
+      "step": 445
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999996484962508,
+      "loss": 2.7622,
+      "step": 450
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999996406411347,
+      "loss": 2.8935,
+      "step": 455
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999999632699216,
+      "loss": 2.9804,
+      "step": 460
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999999624670495,
+      "loss": 2.8613,
+      "step": 465
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999996165549712,
+      "loss": 3.0941,
+      "step": 470
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999996083526445,
+      "loss": 2.6941,
+      "step": 475
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999996000635156,
+      "loss": 2.9191,
+      "step": 480
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999999591687584,
+      "loss": 3.0899,
+      "step": 485
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999995832248497,
+      "loss": 2.9366,
+      "step": 490
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999999574675313,
+      "loss": 2.9312,
+      "step": 495
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999995660389735,
+      "loss": 2.9255,
+      "step": 500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999995573158317,
+      "loss": 2.8753,
+      "step": 505
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999999548505887,
+      "loss": 2.8521,
+      "step": 510
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000199999953960914,
+      "loss": 2.8991,
+      "step": 515
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999995306255902,
+      "loss": 2.8577,
+      "step": 520
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999995215552377,
+      "loss": 2.61,
+      "step": 525
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999995123980827,
+      "loss": 2.8954,
+      "step": 530
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999995031541254,
+      "loss": 2.9163,
+      "step": 535
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999994938233652,
+      "loss": 2.8967,
+      "step": 540
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999994844058024,
+      "loss": 2.7085,
+      "step": 545
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999994749014374,
+      "loss": 3.0516,
+      "step": 550
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999994653102695,
+      "loss": 2.7746,
+      "step": 555
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999999455632299,
+      "loss": 2.742,
+      "step": 560
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999999445867526,
+      "loss": 2.8224,
+      "step": 565
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999994360159505,
+      "loss": 2.9539,
+      "step": 570
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999994260775722,
+      "loss": 2.7028,
+      "step": 575
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999994160523917,
+      "loss": 2.763,
+      "step": 580
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999994059404083,
+      "loss": 2.6874,
+      "step": 585
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999993957416225,
+      "loss": 2.9416,
+      "step": 590
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999999385456034,
+      "loss": 2.7574,
+      "step": 595
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999993750836428,
+      "loss": 2.9096,
+      "step": 600
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999993646244494,
+      "loss": 2.6635,
+      "step": 605
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999999354078453,
+      "loss": 2.8138,
+      "step": 610
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999993434456544,
+      "loss": 2.8623,
+      "step": 615
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999999332726053,
+      "loss": 2.716,
+      "step": 620
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999993219196492,
+      "loss": 2.8355,
+      "step": 625
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999993110264428,
+      "loss": 2.5897,
+      "step": 630
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999993000464336,
+      "loss": 2.574,
+      "step": 635
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999999288979622,
+      "loss": 2.6675,
+      "step": 640
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999999277826008,
+      "loss": 2.7379,
+      "step": 645
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999992665855912,
+      "loss": 2.7875,
+      "step": 650
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999992552583717,
+      "loss": 2.9414,
+      "step": 655
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000199999924384435,
+      "loss": 2.6379,
+      "step": 660
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999992323435257,
+      "loss": 2.9688,
+      "step": 665
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999992207558987,
+      "loss": 2.8554,
+      "step": 670
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999999209081469,
+      "loss": 2.8075,
+      "step": 675
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999999197320237,
+      "loss": 2.8249,
+      "step": 680
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999991854722024,
+      "loss": 2.4325,
+      "step": 685
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999999173537365,
+      "loss": 2.822,
+      "step": 690
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999991615157251,
+      "loss": 2.7508,
+      "step": 695
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999999149407283,
+      "loss": 2.8431,
+      "step": 700
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999999137212038,
+      "loss": 2.7273,
+      "step": 705
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999991249299907,
+      "loss": 2.8935,
+      "step": 710
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999991125611407,
+      "loss": 2.7908,
+      "step": 715
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999991001054884,
+      "loss": 2.7172,
+      "step": 720
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999990875630333,
+      "loss": 2.5546,
+      "step": 725
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999990749337756,
+      "loss": 2.9523,
+      "step": 730
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999990622177153,
+      "loss": 2.7861,
+      "step": 735
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999990494148527,
+      "loss": 2.7639,
+      "step": 740
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999990365251872,
+      "loss": 2.6946,
+      "step": 745
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999990235487194,
+      "loss": 2.8027,
+      "step": 750
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999999010485449,
+      "loss": 2.8132,
+      "step": 755
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999989973353764,
+      "loss": 2.6649,
+      "step": 760
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999989840985006,
+      "loss": 2.805,
+      "step": 765
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999989707748228,
+      "loss": 2.5914,
+      "step": 770
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999989573643424,
+      "loss": 2.6823,
+      "step": 775
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999998943867059,
+      "loss": 2.7104,
+      "step": 780
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999989302829735,
+      "loss": 2.7045,
+      "step": 785
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999989166120857,
+      "loss": 2.7507,
+      "step": 790
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999998902854395,
+      "loss": 2.6759,
+      "step": 795
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999988890099016,
+      "loss": 2.8907,
+      "step": 800
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999988750786057,
+      "loss": 3.0281,
+      "step": 805
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999988610605075,
+      "loss": 2.6338,
+      "step": 810
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999988469556067,
+      "loss": 2.7324,
+      "step": 815
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999988327639036,
+      "loss": 2.7257,
+      "step": 820
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999998818485398,
+      "loss": 2.7441,
+      "step": 825
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999988041200894,
+      "loss": 2.7089,
+      "step": 830
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999987896679786,
+      "loss": 2.6935,
+      "step": 835
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999987751290652,
+      "loss": 2.6191,
+      "step": 840
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999987605033492,
+      "loss": 2.7976,
+      "step": 845
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999987457908306,
+      "loss": 2.7551,
+      "step": 850
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999987309915097,
+      "loss": 2.7361,
+      "step": 855
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999987161053865,
+      "loss": 2.8261,
+      "step": 860
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999987011324604,
+      "loss": 2.8417,
+      "step": 865
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999986860727318,
+      "loss": 2.7262,
+      "step": 870
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999986709262008,
+      "loss": 2.6199,
+      "step": 875
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999986556928673,
+      "loss": 2.8727,
+      "step": 880
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999986403727312,
+      "loss": 2.7327,
+      "step": 885
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999986249657928,
+      "loss": 2.6243,
+      "step": 890
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999986094720518,
+      "loss": 2.7786,
+      "step": 895
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999998593891508,
+      "loss": 2.5695,
+      "step": 900
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999998578224162,
+      "loss": 2.7535,
+      "step": 905
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999985624700136,
+      "loss": 2.9243,
+      "step": 910
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999985466290622,
+      "loss": 2.8599,
+      "step": 915
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999985307013086,
+      "loss": 2.6587,
+      "step": 920
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999985146867526,
+      "loss": 2.7066,
+      "step": 925
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999998498585394,
+      "loss": 2.7567,
+      "step": 930
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999998482397233,
+      "loss": 2.8307,
+      "step": 935
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999984661222696,
+      "loss": 2.6447,
+      "step": 940
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999984497605033,
+      "loss": 2.916,
+      "step": 945
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999984333119345,
+      "loss": 2.8343,
+      "step": 950
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999984167765638,
+      "loss": 2.5178,
+      "step": 955
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000199999840015439,
+      "loss": 2.8607,
+      "step": 960
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999983834454138,
+      "loss": 2.7741,
+      "step": 965
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999983666496357,
+      "loss": 2.8707,
+      "step": 970
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999983497670547,
+      "loss": 2.9699,
+      "step": 975
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999998332797671,
+      "loss": 2.5461,
+      "step": 980
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999998315741485,
+      "loss": 2.7899,
+      "step": 985
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999982985984967,
+      "loss": 2.6976,
+      "step": 990
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999982813687057,
+      "loss": 2.6443,
+      "step": 995
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999998264052112,
+      "loss": 2.7354,
+      "step": 1000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999982466487165,
+      "loss": 2.7049,
+      "step": 1005
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999998229158518,
+      "loss": 2.4594,
+      "step": 1010
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999998211581517,
+      "loss": 2.5669,
+      "step": 1015
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999981939177136,
+      "loss": 2.92,
+      "step": 1020
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999998176167108,
+      "loss": 2.7088,
+      "step": 1025
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999981583297,
+      "loss": 2.6139,
+      "step": 1030
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999998140405489,
+      "loss": 2.5142,
+      "step": 1035
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999998122394476,
+      "loss": 2.7047,
+      "step": 1040
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999981042966602,
+      "loss": 2.6954,
+      "step": 1045
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999980861120422,
+      "loss": 2.8237,
+      "step": 1050
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999980678406213,
+      "loss": 2.7916,
+      "step": 1055
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999980494823984,
+      "loss": 2.7145,
+      "step": 1060
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999998031037373,
+      "loss": 2.8736,
+      "step": 1065
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999998012505545,
+      "loss": 2.656,
+      "step": 1070
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999979938869147,
+      "loss": 2.5206,
+      "step": 1075
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999997975181482,
+      "loss": 2.8481,
+      "step": 1080
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999979563892467,
+      "loss": 2.8695,
+      "step": 1085
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999979375102088,
+      "loss": 2.5611,
+      "step": 1090
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999979185443686,
+      "loss": 2.7148,
+      "step": 1095
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999978994917256,
+      "loss": 2.8026,
+      "step": 1100
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999978803522808,
+      "loss": 2.5893,
+      "step": 1105
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999978611260334,
+      "loss": 2.6701,
+      "step": 1110
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999978418129831,
+      "loss": 2.7724,
+      "step": 1115
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999978224131306,
+      "loss": 2.4461,
+      "step": 1120
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999997802926476,
+      "loss": 2.7933,
+      "step": 1125
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999977833530186,
+      "loss": 2.6923,
+      "step": 1130
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999977636927588,
+      "loss": 2.6523,
+      "step": 1135
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999977439456968,
+      "loss": 2.6436,
+      "step": 1140
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999977241118319,
+      "loss": 2.8382,
+      "step": 1145
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999977041911652,
+      "loss": 2.7091,
+      "step": 1150
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999976841836956,
+      "loss": 2.8385,
+      "step": 1155
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999976640894235,
+      "loss": 2.3165,
+      "step": 1160
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999976439083494,
+      "loss": 2.8163,
+      "step": 1165
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999976236404726,
+      "loss": 2.7067,
+      "step": 1170
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999976032857933,
+      "loss": 2.5457,
+      "step": 1175
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999975828443117,
+      "loss": 2.7439,
+      "step": 1180
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999975623160278,
+      "loss": 2.6707,
+      "step": 1185
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999975417009415,
+      "loss": 2.5346,
+      "step": 1190
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999975209990524,
+      "loss": 2.7652,
+      "step": 1195
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999975002103613,
+      "loss": 2.6012,
+      "step": 1200
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999974793348676,
+      "loss": 2.7155,
+      "step": 1205
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999974583725716,
+      "loss": 2.8679,
+      "step": 1210
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999974373234732,
+      "loss": 2.6568,
+      "step": 1215
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999974161875723,
+      "loss": 2.763,
+      "step": 1220
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999973949648688,
+      "loss": 2.5574,
+      "step": 1225
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999973736553633,
+      "loss": 2.629,
+      "step": 1230
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999973522590554,
+      "loss": 2.5492,
+      "step": 1235
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999997330775945,
+      "loss": 2.4964,
+      "step": 1240
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999997309206032,
+      "loss": 2.8257,
+      "step": 1245
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999972875493167,
+      "loss": 2.9205,
+      "step": 1250
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999997265805799,
+      "loss": 2.5601,
+      "step": 1255
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999997243975479,
+      "loss": 2.6179,
+      "step": 1260
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999972220583568,
+      "loss": 2.5755,
+      "step": 1265
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999972000544317,
+      "loss": 2.4769,
+      "step": 1270
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999971779637046,
+      "loss": 2.6268,
+      "step": 1275
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999971557861749,
+      "loss": 2.4707,
+      "step": 1280
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999997133521843,
+      "loss": 2.6769,
+      "step": 1285
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999971111707088,
+      "loss": 2.514,
+      "step": 1290
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999970887327718,
+      "loss": 2.5784,
+      "step": 1295
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999997066208033,
+      "loss": 2.8341,
+      "step": 1300
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999970435964913,
+      "loss": 2.6296,
+      "step": 1305
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999970208981475,
+      "loss": 2.6278,
+      "step": 1310
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999996998113001,
+      "loss": 2.5416,
+      "step": 1315
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999969752410528,
+      "loss": 2.673,
+      "step": 1320
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999969522823018,
+      "loss": 2.6878,
+      "step": 1325
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999969292367484,
+      "loss": 2.5721,
+      "step": 1330
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999969061043927,
+      "loss": 2.6098,
+      "step": 1335
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999968828852347,
+      "loss": 2.7936,
+      "step": 1340
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999968595792744,
+      "loss": 2.5121,
+      "step": 1345
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999968361865116,
+      "loss": 2.8514,
+      "step": 1350
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999968127069466,
+      "loss": 2.4604,
+      "step": 1355
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999967891405791,
+      "loss": 2.6636,
+      "step": 1360
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999967654874093,
+      "loss": 2.5993,
+      "step": 1365
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999967417474372,
+      "loss": 2.6292,
+      "step": 1370
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999967179206628,
+      "loss": 2.6366,
+      "step": 1375
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999996694007086,
+      "loss": 2.7079,
+      "step": 1380
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999966700067065,
+      "loss": 2.6132,
+      "step": 1385
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999996645919525,
+      "loss": 2.6106,
+      "step": 1390
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999966217455412,
+      "loss": 2.352,
+      "step": 1395
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999965974847552,
+      "loss": 2.439,
+      "step": 1400
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999965731371666,
+      "loss": 2.7842,
+      "step": 1405
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999965487027757,
+      "loss": 2.5492,
+      "step": 1410
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999965241815826,
+      "loss": 2.8821,
+      "step": 1415
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999996499573587,
+      "loss": 2.6805,
+      "step": 1420
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999964748787893,
+      "loss": 2.4762,
+      "step": 1425
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999964500971892,
+      "loss": 2.6549,
+      "step": 1430
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999964252287867,
+      "loss": 2.6132,
+      "step": 1435
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999964002735817,
+      "loss": 2.4747,
+      "step": 1440
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999996375231575,
+      "loss": 2.8531,
+      "step": 1445
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999963501027653,
+      "loss": 2.7625,
+      "step": 1450
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999963248871537,
+      "loss": 2.5995,
+      "step": 1455
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999962995847397,
+      "loss": 2.6859,
+      "step": 1460
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999962741955234,
+      "loss": 2.8208,
+      "step": 1465
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999962487195048,
+      "loss": 2.6235,
+      "step": 1470
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999962231566836,
+      "loss": 2.6238,
+      "step": 1475
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999961975070604,
+      "loss": 2.6362,
+      "step": 1480
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999996171770635,
+      "loss": 2.6904,
+      "step": 1485
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999961459474073,
+      "loss": 2.5746,
+      "step": 1490
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999961200373772,
+      "loss": 2.7593,
+      "step": 1495
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999960940405448,
+      "loss": 2.3081,
+      "step": 1500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000199999606795691,
+      "loss": 2.8279,
+      "step": 1505
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999996041786473,
+      "loss": 2.6013,
+      "step": 1510
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999960155292338,
+      "loss": 2.7186,
+      "step": 1515
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999959891851924,
+      "loss": 2.3538,
+      "step": 1520
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999959627543486,
+      "loss": 2.6952,
+      "step": 1525
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999959362367023,
+      "loss": 2.4529,
+      "step": 1530
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999959096322537,
+      "loss": 2.8643,
+      "step": 1535
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999958829410033,
+      "loss": 2.6397,
+      "step": 1540
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000199999585616295,
+      "loss": 2.7624,
+      "step": 1545
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999995829298095,
+      "loss": 2.4717,
+      "step": 1550
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999958023464375,
+      "loss": 2.9581,
+      "step": 1555
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999957753079778,
+      "loss": 2.5924,
+      "step": 1560
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999957481827162,
+      "loss": 2.8503,
+      "step": 1565
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999957209706516,
+      "loss": 2.7108,
+      "step": 1570
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999995693671785,
+      "loss": 2.7926,
+      "step": 1575
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999956662861165,
+      "loss": 2.6825,
+      "step": 1580
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999956388136453,
+      "loss": 2.4061,
+      "step": 1585
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999956112543723,
+      "loss": 2.8156,
+      "step": 1590
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999955836082965,
+      "loss": 2.6561,
+      "step": 1595
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999995555875419,
+      "loss": 2.6375,
+      "step": 1600
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999995528055739,
+      "loss": 2.5209,
+      "step": 1605
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999955001492566,
+      "loss": 2.5621,
+      "step": 1610
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999954721559724,
+      "loss": 2.7921,
+      "step": 1615
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999954440758856,
+      "loss": 2.7439,
+      "step": 1620
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999954159089967,
+      "loss": 2.6912,
+      "step": 1625
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999953876553056,
+      "loss": 2.4257,
+      "step": 1630
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999995359314812,
+      "loss": 2.566,
+      "step": 1635
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999953308875166,
+      "loss": 2.687,
+      "step": 1640
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999953023734185,
+      "loss": 2.8141,
+      "step": 1645
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999952737725186,
+      "loss": 2.6272,
+      "step": 1650
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999952450848162,
+      "loss": 2.539,
+      "step": 1655
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999952163103117,
+      "loss": 2.3768,
+      "step": 1660
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999951874490052,
+      "loss": 2.5902,
+      "step": 1665
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999951585008963,
+      "loss": 2.5818,
+      "step": 1670
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999951294659852,
+      "loss": 2.6016,
+      "step": 1675
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999951003442718,
+      "loss": 2.843,
+      "step": 1680
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999950711357563,
+      "loss": 2.7841,
+      "step": 1685
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999950418404385,
+      "loss": 2.4404,
+      "step": 1690
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999950124583186,
+      "loss": 2.7421,
+      "step": 1695
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999949829893968,
+      "loss": 2.9662,
+      "step": 1700
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999949534336723,
+      "loss": 2.5873,
+      "step": 1705
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999994923791146,
+      "loss": 2.5671,
+      "step": 1710
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999948940618173,
+      "loss": 2.5707,
+      "step": 1715
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999948642456861,
+      "loss": 2.7794,
+      "step": 1720
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999948343427533,
+      "loss": 2.7436,
+      "step": 1725
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999994804353018,
+      "loss": 2.8675,
+      "step": 1730
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999947742764803,
+      "loss": 2.6696,
+      "step": 1735
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999994744113141,
+      "loss": 2.7555,
+      "step": 1740
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999947138629992,
+      "loss": 2.6759,
+      "step": 1745
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999946835260555,
+      "loss": 2.5621,
+      "step": 1750
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999946531023093,
+      "loss": 2.7678,
+      "step": 1755
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999994622591761,
+      "loss": 2.6351,
+      "step": 1760
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999945919944105,
+      "loss": 2.6966,
+      "step": 1765
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999994561310258,
+      "loss": 2.6475,
+      "step": 1770
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999945305393033,
+      "loss": 2.7657,
+      "step": 1775
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999944996815466,
+      "loss": 2.3656,
+      "step": 1780
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999944687369874,
+      "loss": 2.7185,
+      "step": 1785
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999994437705626,
+      "loss": 2.5218,
+      "step": 1790
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999994406587463,
+      "loss": 2.6173,
+      "step": 1795
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999943753824974,
+      "loss": 2.6548,
+      "step": 1800
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999943440907297,
+      "loss": 2.8232,
+      "step": 1805
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000199999431271216,
+      "loss": 2.6127,
+      "step": 1810
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999942812467882,
+      "loss": 2.5608,
+      "step": 1815
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999942496946142,
+      "loss": 2.869,
+      "step": 1820
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999994218055638,
+      "loss": 2.442,
+      "step": 1825
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000199999418632986,
+      "loss": 2.4997,
+      "step": 1830
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999941545172795,
+      "loss": 2.4263,
+      "step": 1835
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999994122617897,
+      "loss": 2.6923,
+      "step": 1840
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999940906317124,
+      "loss": 2.479,
+      "step": 1845
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999940585587256,
+      "loss": 2.6618,
+      "step": 1850
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999994026398937,
+      "loss": 2.8523,
+      "step": 1855
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999939941523458,
+      "loss": 2.6514,
+      "step": 1860
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999939618189528,
+      "loss": 2.4561,
+      "step": 1865
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999939293987575,
+      "loss": 2.8679,
+      "step": 1870
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999938968917605,
+      "loss": 2.81,
+      "step": 1875
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999938642979611,
+      "loss": 2.7132,
+      "step": 1880
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999938316173598,
+      "loss": 2.6953,
+      "step": 1885
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999993798849956,
+      "loss": 2.5594,
+      "step": 1890
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999937659957503,
+      "loss": 2.8072,
+      "step": 1895
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999937330547428,
+      "loss": 2.7259,
+      "step": 1900
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999993700026933,
+      "loss": 2.602,
+      "step": 1905
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999993666912321,
+      "loss": 2.6015,
+      "step": 1910
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999993633710907,
+      "loss": 2.7177,
+      "step": 1915
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999993600422691,
+      "loss": 2.7859,
+      "step": 1920
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999935670476728,
+      "loss": 2.4358,
+      "step": 1925
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999935335858528,
+      "loss": 2.5998,
+      "step": 1930
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999935000372305,
+      "loss": 2.7423,
+      "step": 1935
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999993466401806,
+      "loss": 2.5114,
+      "step": 1940
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999934326795797,
+      "loss": 2.4934,
+      "step": 1945
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999933988705513,
+      "loss": 2.6945,
+      "step": 1950
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999933649747208,
+      "loss": 2.6669,
+      "step": 1955
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999933309920885,
+      "loss": 2.787,
+      "step": 1960
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999932969226537,
+      "loss": 2.6692,
+      "step": 1965
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999993262766417,
+      "loss": 2.6079,
+      "step": 1970
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999932285233785,
+      "loss": 2.4989,
+      "step": 1975
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999931941935378,
+      "loss": 2.774,
+      "step": 1980
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999993159776895,
+      "loss": 2.6556,
+      "step": 1985
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999931252734503,
+      "loss": 2.5328,
+      "step": 1990
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999930906832033,
+      "loss": 2.6393,
+      "step": 1995
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999930560061547,
+      "loss": 2.6058,
+      "step": 2000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999930212423036,
+      "loss": 2.8369,
+      "step": 2005
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999992986391651,
+      "loss": 2.64,
+      "step": 2010
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999992951454196,
+      "loss": 2.6686,
+      "step": 2015
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999992916429939,
+      "loss": 2.9274,
+      "step": 2020
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000199999288131888,
+      "loss": 2.4095,
+      "step": 2025
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999928461210192,
+      "loss": 2.5919,
+      "step": 2030
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999992810836356,
+      "loss": 2.6322,
+      "step": 2035
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999927754648912,
+      "loss": 2.587,
+      "step": 2040
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999927400066243,
+      "loss": 2.6958,
+      "step": 2045
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999927044615554,
+      "loss": 2.5001,
+      "step": 2050
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999926688296844,
+      "loss": 2.7739,
+      "step": 2055
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999926331110116,
+      "loss": 2.6311,
+      "step": 2060
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999925973055365,
+      "loss": 2.2464,
+      "step": 2065
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999925614132597,
+      "loss": 2.6054,
+      "step": 2070
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999925254341806,
+      "loss": 2.7537,
+      "step": 2075
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999924893683,
+      "loss": 2.674,
+      "step": 2080
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999992453215617,
+      "loss": 2.7459,
+      "step": 2085
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999924169761322,
+      "loss": 2.6581,
+      "step": 2090
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999923806498452,
+      "loss": 2.5208,
+      "step": 2095
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999923442367567,
+      "loss": 2.9183,
+      "step": 2100
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999992307736866,
+      "loss": 2.4701,
+      "step": 2105
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999922711501732,
+      "loss": 2.6732,
+      "step": 2110
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999922344766786,
+      "loss": 2.6728,
+      "step": 2115
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999992197716382,
+      "loss": 2.561,
+      "step": 2120
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999921608692835,
+      "loss": 2.4256,
+      "step": 2125
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999921239353832,
+      "loss": 2.5284,
+      "step": 2130
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999920869146807,
+      "loss": 2.7438,
+      "step": 2135
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999920498071761,
+      "loss": 2.5915,
+      "step": 2140
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000199999201261287,
+      "loss": 2.6054,
+      "step": 2145
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999919753317617,
+      "loss": 2.8368,
+      "step": 2150
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999919379638516,
+      "loss": 2.5809,
+      "step": 2155
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999919005091394,
+      "loss": 2.6437,
+      "step": 2160
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999918629676254,
+      "loss": 2.611,
+      "step": 2165
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999918253393094,
+      "loss": 2.516,
+      "step": 2170
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999917876241917,
+      "loss": 2.3178,
+      "step": 2175
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999991749822272,
+      "loss": 2.5397,
+      "step": 2180
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000199999171193355,
+      "loss": 2.6697,
+      "step": 2185
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999916739580264,
+      "loss": 2.5902,
+      "step": 2190
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999991635895701,
+      "loss": 2.6564,
+      "step": 2195
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999915977465737,
+      "loss": 2.514,
+      "step": 2200
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999915595106442,
+      "loss": 2.6288,
+      "step": 2205
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999991521187913,
+      "loss": 2.7036,
+      "step": 2210
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000199999148277838,
+      "loss": 2.6969,
+      "step": 2215
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999991444282045,
+      "loss": 2.4181,
+      "step": 2220
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999914056989082,
+      "loss": 2.5399,
+      "step": 2225
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999913670289696,
+      "loss": 2.3588,
+      "step": 2230
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999913282722287,
+      "loss": 2.7133,
+      "step": 2235
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999912894286864,
+      "loss": 2.5371,
+      "step": 2240
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999991250498342,
+      "loss": 2.8503,
+      "step": 2245
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999912114811958,
+      "loss": 2.4967,
+      "step": 2250
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999911723772478,
+      "loss": 2.6704,
+      "step": 2255
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999911331864976,
+      "loss": 2.5961,
+      "step": 2260
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999910939089458,
+      "loss": 2.5097,
+      "step": 2265
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999991054544592,
+      "loss": 2.598,
+      "step": 2270
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999910150934368,
+      "loss": 2.7947,
+      "step": 2275
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999909755554794,
+      "loss": 2.4228,
+      "step": 2280
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000199999093593072,
+      "loss": 2.5667,
+      "step": 2285
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999990896219159,
+      "loss": 2.5555,
+      "step": 2290
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999990856420796,
+      "loss": 2.5606,
+      "step": 2295
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999908165356314,
+      "loss": 2.6759,
+      "step": 2300
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999990776563665,
+      "loss": 2.8376,
+      "step": 2305
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999907365048965,
+      "loss": 2.5424,
+      "step": 2310
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999906963593262,
+      "loss": 2.4368,
+      "step": 2315
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999906561269542,
+      "loss": 2.6155,
+      "step": 2320
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999906158077804,
+      "loss": 2.5441,
+      "step": 2325
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999905754018046,
+      "loss": 2.5587,
+      "step": 2330
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999905349090273,
+      "loss": 2.5802,
+      "step": 2335
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999904943294482,
+      "loss": 2.6721,
+      "step": 2340
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999904536630668,
+      "loss": 2.5217,
+      "step": 2345
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999904129098842,
+      "loss": 2.6426,
+      "step": 2350
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999903720698992,
+      "loss": 2.4764,
+      "step": 2355
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999990331143113,
+      "loss": 2.6398,
+      "step": 2360
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999902901295243,
+      "loss": 2.8396,
+      "step": 2365
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999902490291347,
+      "loss": 2.6032,
+      "step": 2370
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999990207841943,
+      "loss": 2.7985,
+      "step": 2375
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999901665679492,
+      "loss": 2.7499,
+      "step": 2380
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999990125207154,
+      "loss": 2.7408,
+      "step": 2385
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999900837595566,
+      "loss": 2.5313,
+      "step": 2390
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999900422251575,
+      "loss": 2.3107,
+      "step": 2395
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999990000603957,
+      "loss": 2.7134,
+      "step": 2400
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999899588959546,
+      "loss": 2.4859,
+      "step": 2405
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999899171011505,
+      "loss": 2.6712,
+      "step": 2410
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999898752195446,
+      "loss": 2.8095,
+      "step": 2415
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999989833251137,
+      "loss": 2.5598,
+      "step": 2420
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999897911959273,
+      "loss": 2.4414,
+      "step": 2425
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999989749053916,
+      "loss": 2.5381,
+      "step": 2430
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999897068251031,
+      "loss": 2.5387,
+      "step": 2435
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999896645094887,
+      "loss": 2.607,
+      "step": 2440
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999896221070725,
+      "loss": 2.8016,
+      "step": 2445
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999989579617854,
+      "loss": 2.7545,
+      "step": 2450
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999895370418345,
+      "loss": 2.7226,
+      "step": 2455
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999894943790127,
+      "loss": 2.7263,
+      "step": 2460
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999894516293897,
+      "loss": 2.714,
+      "step": 2465
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999894087929646,
+      "loss": 2.8175,
+      "step": 2470
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999989365869738,
+      "loss": 2.5644,
+      "step": 2475
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999893228597095,
+      "loss": 2.4785,
+      "step": 2480
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999892797628794,
+      "loss": 2.6025,
+      "step": 2485
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999892365792475,
+      "loss": 2.7969,
+      "step": 2490
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999891933088142,
+      "loss": 2.7892,
+      "step": 2495
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999989149951579,
+      "loss": 2.556,
+      "step": 2500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999891065075422,
+      "loss": 2.5106,
+      "step": 2505
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999890629767038,
+      "loss": 2.5166,
+      "step": 2510
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999890193590634,
+      "loss": 2.5847,
+      "step": 2515
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999889756546215,
+      "loss": 2.6806,
+      "step": 2520
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999988931863378,
+      "loss": 2.6937,
+      "step": 2525
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999888879853326,
+      "loss": 2.6377,
+      "step": 2530
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999888440204856,
+      "loss": 2.3986,
+      "step": 2535
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999887999688372,
+      "loss": 2.8614,
+      "step": 2540
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999887558303872,
+      "loss": 2.8034,
+      "step": 2545
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999887116051352,
+      "loss": 2.6652,
+      "step": 2550
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999886672930818,
+      "loss": 2.703,
+      "step": 2555
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999886228942265,
+      "loss": 2.5679,
+      "step": 2560
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999885784085698,
+      "loss": 2.3774,
+      "step": 2565
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999885338361115,
+      "loss": 2.7305,
+      "step": 2570
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999884891768512,
+      "loss": 2.1613,
+      "step": 2575
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999884444307897,
+      "loss": 2.6024,
+      "step": 2580
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999883995979264,
+      "loss": 2.5547,
+      "step": 2585
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999883546782614,
+      "loss": 2.6603,
+      "step": 2590
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999883096717948,
+      "loss": 2.3867,
+      "step": 2595
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999882645785268,
+      "loss": 2.5352,
+      "step": 2600
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999988219398457,
+      "loss": 2.5642,
+      "step": 2605
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999881741315857,
+      "loss": 2.86,
+      "step": 2610
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999881287779126,
+      "loss": 2.7749,
+      "step": 2615
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999988083337438,
+      "loss": 2.3401,
+      "step": 2620
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999988037810162,
+      "loss": 2.4999,
+      "step": 2625
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999879921960841,
+      "loss": 2.6999,
+      "step": 2630
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999879464952048,
+      "loss": 2.6279,
+      "step": 2635
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999879007075237,
+      "loss": 2.5854,
+      "step": 2640
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999878548330414,
+      "loss": 2.6713,
+      "step": 2645
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999878088717573,
+      "loss": 2.8,
+      "step": 2650
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999987762823672,
+      "loss": 2.4333,
+      "step": 2655
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999877166887843,
+      "loss": 2.8168,
+      "step": 2660
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999876704670957,
+      "loss": 2.467,
+      "step": 2665
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999876241586054,
+      "loss": 2.5969,
+      "step": 2670
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999875777633135,
+      "loss": 2.5396,
+      "step": 2675
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000199998753128122,
+      "loss": 2.5875,
+      "step": 2680
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999987484712325,
+      "loss": 2.5898,
+      "step": 2685
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999874380566285,
+      "loss": 2.7047,
+      "step": 2690
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999873913141306,
+      "loss": 2.5165,
+      "step": 2695
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999987344484831,
+      "loss": 2.5659,
+      "step": 2700
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000199998729756873,
+      "loss": 2.696,
+      "step": 2705
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999872505658273,
+      "loss": 2.474,
+      "step": 2710
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999872034761232,
+      "loss": 2.5292,
+      "step": 2715
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999871562996173,
+      "loss": 2.8167,
+      "step": 2720
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999871090363103,
+      "loss": 2.6075,
+      "step": 2725
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999870616862017,
+      "loss": 2.3508,
+      "step": 2730
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999870142492913,
+      "loss": 2.6551,
+      "step": 2735
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999869667255795,
+      "loss": 2.3473,
+      "step": 2740
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999869191150664,
+      "loss": 2.5731,
+      "step": 2745
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999868714177515,
+      "loss": 2.5612,
+      "step": 2750
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999868236336355,
+      "loss": 2.4575,
+      "step": 2755
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999986775762718,
+      "loss": 2.6937,
+      "step": 2760
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999867278049988,
+      "loss": 2.7548,
+      "step": 2765
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999986679760478,
+      "loss": 2.7528,
+      "step": 2770
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999986631629156,
+      "loss": 2.552,
+      "step": 2775
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999865834110326,
+      "loss": 2.6023,
+      "step": 2780
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999865351061076,
+      "loss": 2.7166,
+      "step": 2785
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999864867143813,
+      "loss": 2.5235,
+      "step": 2790
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999986438235853,
+      "loss": 2.5894,
+      "step": 2795
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999863896705237,
+      "loss": 2.625,
+      "step": 2800
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999986341018393,
+      "loss": 2.506,
+      "step": 2805
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999862922794607,
+      "loss": 2.8826,
+      "step": 2810
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999986243453727,
+      "loss": 2.3339,
+      "step": 2815
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999986194541192,
+      "loss": 2.7514,
+      "step": 2820
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999861455418554,
+      "loss": 2.6697,
+      "step": 2825
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999860964557171,
+      "loss": 2.6295,
+      "step": 2830
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999986047282778,
+      "loss": 2.7592,
+      "step": 2835
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999985998023037,
+      "loss": 2.6853,
+      "step": 2840
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999859486764948,
+      "loss": 2.4563,
+      "step": 2845
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999985899243151,
+      "loss": 2.5586,
+      "step": 2850
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999858497230058,
+      "loss": 2.4144,
+      "step": 2855
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999858001160597,
+      "loss": 2.5669,
+      "step": 2860
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999857504223115,
+      "loss": 2.5779,
+      "step": 2865
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999857006417626,
+      "loss": 2.5671,
+      "step": 2870
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999985650774412,
+      "loss": 2.6281,
+      "step": 2875
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000199998560082026,
+      "loss": 2.7483,
+      "step": 2880
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999855507793065,
+      "loss": 2.496,
+      "step": 2885
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999985500651552,
+      "loss": 2.7721,
+      "step": 2890
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999854504369955,
+      "loss": 2.463,
+      "step": 2895
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999854001356382,
+      "loss": 2.4655,
+      "step": 2900
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999985349747479,
+      "loss": 2.6399,
+      "step": 2905
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999985299272519,
+      "loss": 2.8935,
+      "step": 2910
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999852487107575,
+      "loss": 2.6516,
+      "step": 2915
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999851980621945,
+      "loss": 2.7925,
+      "step": 2920
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999851473268302,
+      "loss": 2.5695,
+      "step": 2925
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999850965046647,
+      "loss": 2.2234,
+      "step": 2930
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999850455956978,
+      "loss": 2.7042,
+      "step": 2935
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999849945999295,
+      "loss": 2.5192,
+      "step": 2940
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999849435173598,
+      "loss": 2.6676,
+      "step": 2945
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999848923479892,
+      "loss": 2.5299,
+      "step": 2950
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999848410918168,
+      "loss": 2.5762,
+      "step": 2955
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999847897488434,
+      "loss": 2.6976,
+      "step": 2960
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999847383190688,
+      "loss": 2.5738,
+      "step": 2965
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999846868024924,
+      "loss": 2.5577,
+      "step": 2970
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999984635199115,
+      "loss": 2.7278,
+      "step": 2975
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999845835089365,
+      "loss": 2.6825,
+      "step": 2980
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999845317319564,
+      "loss": 2.694,
+      "step": 2985
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999844798681752,
+      "loss": 2.4306,
+      "step": 2990
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999844279175924,
+      "loss": 2.6339,
+      "step": 2995
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999843758802084,
+      "loss": 2.3934,
+      "step": 3000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999843237560234,
+      "loss": 2.5188,
+      "step": 3005
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999984271545037,
+      "loss": 2.7121,
+      "step": 3010
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999842192472493,
+      "loss": 2.6028,
+      "step": 3015
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000199998416686266,
+      "loss": 2.5672,
+      "step": 3020
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999841143912702,
+      "loss": 2.6899,
+      "step": 3025
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999984061833079,
+      "loss": 2.509,
+      "step": 3030
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999840091880863,
+      "loss": 2.5404,
+      "step": 3035
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999839564562922,
+      "loss": 2.6446,
+      "step": 3040
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999839036376972,
+      "loss": 2.6019,
+      "step": 3045
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999838507323006,
+      "loss": 2.6164,
+      "step": 3050
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999983797740103,
+      "loss": 2.4493,
+      "step": 3055
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999837446611041,
+      "loss": 2.7621,
+      "step": 3060
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999836914953042,
+      "loss": 2.6179,
+      "step": 3065
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999983638242703,
+      "loss": 2.8092,
+      "step": 3070
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999835849033006,
+      "loss": 2.4296,
+      "step": 3075
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999983531477097,
+      "loss": 2.5016,
+      "step": 3080
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999983477964092,
+      "loss": 2.5551,
+      "step": 3085
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999834243642858,
+      "loss": 2.7123,
+      "step": 3090
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999833706776788,
+      "loss": 2.7107,
+      "step": 3095
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999833169042703,
+      "loss": 2.5635,
+      "step": 3100
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999832630440608,
+      "loss": 2.4079,
+      "step": 3105
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000199998320909705,
+      "loss": 2.6807,
+      "step": 3110
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999831550632382,
+      "loss": 2.6134,
+      "step": 3115
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999831009426247,
+      "loss": 2.4098,
+      "step": 3120
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999830467352106,
+      "loss": 2.3827,
+      "step": 3125
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999829924409953,
+      "loss": 2.6387,
+      "step": 3130
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999829380599787,
+      "loss": 2.261,
+      "step": 3135
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999982883592161,
+      "loss": 2.6373,
+      "step": 3140
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999982829037542,
+      "loss": 2.4054,
+      "step": 3145
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999827743961222,
+      "loss": 2.5333,
+      "step": 3150
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999982719667901,
+      "loss": 2.4284,
+      "step": 3155
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999982664852879,
+      "loss": 2.2998,
+      "step": 3160
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999826099510552,
+      "loss": 2.4798,
+      "step": 3165
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999982554962431,
+      "loss": 2.7227,
+      "step": 3170
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999824998870054,
+      "loss": 2.5127,
+      "step": 3175
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999824447247786,
+      "loss": 2.5939,
+      "step": 3180
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999823894757506,
+      "loss": 2.5919,
+      "step": 3185
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999982334139922,
+      "loss": 2.5951,
+      "step": 3190
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999822787172917,
+      "loss": 2.7247,
+      "step": 3195
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999822232078605,
+      "loss": 2.6713,
+      "step": 3200
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999821676116284,
+      "loss": 2.4816,
+      "step": 3205
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999982111928595,
+      "loss": 2.5262,
+      "step": 3210
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999820561587608,
+      "loss": 2.7087,
+      "step": 3215
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999820003021253,
+      "loss": 2.8628,
+      "step": 3220
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999819443586889,
+      "loss": 2.6129,
+      "step": 3225
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999818883284514,
+      "loss": 2.3413,
+      "step": 3230
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999818322114128,
+      "loss": 2.6626,
+      "step": 3235
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999981776007573,
+      "loss": 2.6738,
+      "step": 3240
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999817197169324,
+      "loss": 2.5838,
+      "step": 3245
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999816633394906,
+      "loss": 2.5262,
+      "step": 3250
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999981606875248,
+      "loss": 2.852,
+      "step": 3255
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999815503242042,
+      "loss": 2.4902,
+      "step": 3260
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999814936863593,
+      "loss": 2.6218,
+      "step": 3265
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999814369617137,
+      "loss": 2.6053,
+      "step": 3270
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999813801502666,
+      "loss": 2.6452,
+      "step": 3275
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999813232520185,
+      "loss": 2.7253,
+      "step": 3280
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999812662669698,
+      "loss": 2.4042,
+      "step": 3285
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000199998120919512,
+      "loss": 2.531,
+      "step": 3290
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999811520364692,
+      "loss": 2.5915,
+      "step": 3295
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999981094791017,
+      "loss": 2.7373,
+      "step": 3300
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999810374587642,
+      "loss": 2.6961,
+      "step": 3305
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999809800397105,
+      "loss": 2.6234,
+      "step": 3310
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999809225338555,
+      "loss": 2.7268,
+      "step": 3315
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999808649412,
+      "loss": 2.5442,
+      "step": 3320
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999980807261743,
+      "loss": 2.489,
+      "step": 3325
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999807494954855,
+      "loss": 2.5636,
+      "step": 3330
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999980691642427,
+      "loss": 2.3505,
+      "step": 3335
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999980633702567,
+      "loss": 2.2822,
+      "step": 3340
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999805756759065,
+      "loss": 2.5858,
+      "step": 3345
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999805175624448,
+      "loss": 2.5202,
+      "step": 3350
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999804593621824,
+      "loss": 2.5763,
+      "step": 3355
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999804010751193,
+      "loss": 2.3721,
+      "step": 3360
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999803427012548,
+      "loss": 2.8356,
+      "step": 3365
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999802842405898,
+      "loss": 2.676,
+      "step": 3370
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999802256931234,
+      "loss": 2.6306,
+      "step": 3375
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999801670588565,
+      "loss": 2.5209,
+      "step": 3380
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999801083377884,
+      "loss": 2.4824,
+      "step": 3385
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999800495299196,
+      "loss": 2.8387,
+      "step": 3390
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999799906352496,
+      "loss": 2.8534,
+      "step": 3395
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999799316537792,
+      "loss": 2.3639,
+      "step": 3400
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999798725855078,
+      "loss": 2.6524,
+      "step": 3405
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999798134304352,
+      "loss": 2.8767,
+      "step": 3410
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999979754188562,
+      "loss": 2.4607,
+      "step": 3415
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999796948598877,
+      "loss": 2.6159,
+      "step": 3420
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999796354444127,
+      "loss": 2.5035,
+      "step": 3425
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999795759421368,
+      "loss": 2.5866,
+      "step": 3430
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000199997951635306,
+      "loss": 2.5947,
+      "step": 3435
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999794566771825,
+      "loss": 2.6712,
+      "step": 3440
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999793969145043,
+      "loss": 2.7967,
+      "step": 3445
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999793370650249,
+      "loss": 2.4733,
+      "step": 3450
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999792771287447,
+      "loss": 2.5811,
+      "step": 3455
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999792171056637,
+      "loss": 2.7599,
+      "step": 3460
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999979156995782,
+      "loss": 2.4645,
+      "step": 3465
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999790967990995,
+      "loss": 2.7224,
+      "step": 3470
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999979036515616,
+      "loss": 2.5087,
+      "step": 3475
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999789761453317,
+      "loss": 2.8189,
+      "step": 3480
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999789156882467,
+      "loss": 2.5146,
+      "step": 3485
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999978855144361,
+      "loss": 2.7451,
+      "step": 3490
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999787945136746,
+      "loss": 2.6152,
+      "step": 3495
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999978733796187,
+      "loss": 2.399,
+      "step": 3500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999786729918987,
+      "loss": 2.5335,
+      "step": 3505
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000199997861210081,
+      "loss": 2.4403,
+      "step": 3510
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999785511229203,
+      "loss": 2.6851,
+      "step": 3515
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999784900582296,
+      "loss": 2.7007,
+      "step": 3520
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999784289067386,
+      "loss": 2.4623,
+      "step": 3525
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999783676684466,
+      "loss": 2.5301,
+      "step": 3530
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999978306343354,
+      "loss": 2.2027,
+      "step": 3535
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000199997824493146,
+      "loss": 2.6527,
+      "step": 3540
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999978183432766,
+      "loss": 2.631,
+      "step": 3545
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999978121847271,
+      "loss": 2.5573,
+      "step": 3550
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999780601749753,
+      "loss": 2.4258,
+      "step": 3555
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999977998415879,
+      "loss": 2.6852,
+      "step": 3560
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999977936569982,
+      "loss": 2.5886,
+      "step": 3565
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999778746372838,
+      "loss": 2.6689,
+      "step": 3570
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999778126177854,
+      "loss": 2.6879,
+      "step": 3575
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999777505114863,
+      "loss": 2.6052,
+      "step": 3580
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999776883183862,
+      "loss": 2.5364,
+      "step": 3585
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999776260384858,
+      "loss": 2.7237,
+      "step": 3590
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999775636717843,
+      "loss": 2.6191,
+      "step": 3595
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999775012182825,
+      "loss": 2.6068,
+      "step": 3600
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999774386779797,
+      "loss": 2.4453,
+      "step": 3605
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999773760508762,
+      "loss": 2.6097,
+      "step": 3610
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999773133369723,
+      "loss": 2.4836,
+      "step": 3615
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999772505362677,
+      "loss": 2.7328,
+      "step": 3620
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999771876487622,
+      "loss": 2.6265,
+      "step": 3625
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999771246744565,
+      "loss": 2.5548,
+      "step": 3630
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999770616133496,
+      "loss": 2.4861,
+      "step": 3635
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999769984654426,
+      "loss": 2.6367,
+      "step": 3640
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999769352307346,
+      "loss": 2.5729,
+      "step": 3645
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999976871909226,
+      "loss": 2.6693,
+      "step": 3650
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999976808500917,
+      "loss": 2.6048,
+      "step": 3655
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999976745005807,
+      "loss": 2.5641,
+      "step": 3660
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999766814238968,
+      "loss": 2.5885,
+      "step": 3665
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999976617755186,
+      "loss": 2.7146,
+      "step": 3670
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999765539996744,
+      "loss": 2.632,
+      "step": 3675
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999764901573624,
+      "loss": 2.4664,
+      "step": 3680
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999764262282495,
+      "loss": 2.2652,
+      "step": 3685
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999763622123362,
+      "loss": 2.5475,
+      "step": 3690
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999762981096224,
+      "loss": 2.5403,
+      "step": 3695
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999976233920108,
+      "loss": 2.8656,
+      "step": 3700
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999976169643793,
+      "loss": 2.6758,
+      "step": 3705
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999761052806773,
+      "loss": 2.4518,
+      "step": 3710
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999760408307614,
+      "loss": 2.5698,
+      "step": 3715
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999759762940445,
+      "loss": 2.5766,
+      "step": 3720
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999759116705272,
+      "loss": 2.6878,
+      "step": 3725
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999758469602097,
+      "loss": 2.4704,
+      "step": 3730
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999757821630913,
+      "loss": 2.56,
+      "step": 3735
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999757172791725,
+      "loss": 2.4563,
+      "step": 3740
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999756523084533,
+      "loss": 2.2081,
+      "step": 3745
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999755872509333,
+      "loss": 2.6101,
+      "step": 3750
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999755221066127,
+      "loss": 2.5998,
+      "step": 3755
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999975456875492,
+      "loss": 2.4085,
+      "step": 3760
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999753915575706,
+      "loss": 2.3679,
+      "step": 3765
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999975326152849,
+      "loss": 2.7311,
+      "step": 3770
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999752606613265,
+      "loss": 2.6587,
+      "step": 3775
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999751950830038,
+      "loss": 2.4725,
+      "step": 3780
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999751294178805,
+      "loss": 2.5657,
+      "step": 3785
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999750636659567,
+      "loss": 2.4418,
+      "step": 3790
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999749978272325,
+      "loss": 2.4117,
+      "step": 3795
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999974931901708,
+      "loss": 2.6565,
+      "step": 3800
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999748658893827,
+      "loss": 2.6,
+      "step": 3805
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999747997902573,
+      "loss": 2.4808,
+      "step": 3810
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999747336043312,
+      "loss": 2.4266,
+      "step": 3815
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999746673316046,
+      "loss": 2.5674,
+      "step": 3820
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999746009720777,
+      "loss": 2.6225,
+      "step": 3825
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999745345257504,
+      "loss": 2.7522,
+      "step": 3830
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999744679926226,
+      "loss": 2.4364,
+      "step": 3835
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999744013726947,
+      "loss": 2.5721,
+      "step": 3840
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999743346659662,
+      "loss": 2.4399,
+      "step": 3845
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999742678724372,
+      "loss": 2.671,
+      "step": 3850
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999742009921078,
+      "loss": 2.4909,
+      "step": 3855
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999974134024978,
+      "loss": 2.5082,
+      "step": 3860
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999740669710478,
+      "loss": 2.375,
+      "step": 3865
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999739998303177,
+      "loss": 2.6448,
+      "step": 3870
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999739326027866,
+      "loss": 2.5758,
+      "step": 3875
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999738652884554,
+      "loss": 2.512,
+      "step": 3880
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999737978873238,
+      "loss": 2.4797,
+      "step": 3885
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999973730399392,
+      "loss": 2.4649,
+      "step": 3890
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999736628246594,
+      "loss": 2.6255,
+      "step": 3895
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999973595163127,
+      "loss": 2.2548,
+      "step": 3900
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999735274147938,
+      "loss": 2.485,
+      "step": 3905
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999973459579661,
+      "loss": 2.5685,
+      "step": 3910
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999973391657727,
+      "loss": 2.6105,
+      "step": 3915
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999733236489932,
+      "loss": 2.7149,
+      "step": 3920
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999973255553459,
+      "loss": 2.6762,
+      "step": 3925
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999731873711246,
+      "loss": 2.3135,
+      "step": 3930
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999731191019897,
+      "loss": 2.4766,
+      "step": 3935
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999730507460546,
+      "loss": 2.6979,
+      "step": 3940
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999972982303319,
+      "loss": 2.3291,
+      "step": 3945
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999729137737835,
+      "loss": 2.4732,
+      "step": 3950
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999728451574472,
+      "loss": 2.8404,
+      "step": 3955
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999727764543113,
+      "loss": 2.4579,
+      "step": 3960
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999727076643747,
+      "loss": 2.5311,
+      "step": 3965
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999726387876377,
+      "loss": 2.677,
+      "step": 3970
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999725698241009,
+      "loss": 2.7106,
+      "step": 3975
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999725007737636,
+      "loss": 2.4307,
+      "step": 3980
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999724316366262,
+      "loss": 2.5786,
+      "step": 3985
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999723624126886,
+      "loss": 2.4742,
+      "step": 3990
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999722931019504,
+      "loss": 2.7372,
+      "step": 3995
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999972223704412,
+      "loss": 2.4513,
+      "step": 4000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999721542200737,
+      "loss": 2.5488,
+      "step": 4005
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999972084648935,
+      "loss": 2.5753,
+      "step": 4010
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999720149909962,
+      "loss": 2.6319,
+      "step": 4015
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999719452462573,
+      "loss": 2.5538,
+      "step": 4020
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999718754147182,
+      "loss": 2.4668,
+      "step": 4025
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999718054963787,
+      "loss": 2.6052,
+      "step": 4030
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999971735491239,
+      "loss": 2.3874,
+      "step": 4035
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999716653992992,
+      "loss": 2.62,
+      "step": 4040
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999715952205593,
+      "loss": 2.6644,
+      "step": 4045
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999715249550192,
+      "loss": 2.4428,
+      "step": 4050
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999971454602679,
+      "loss": 2.6147,
+      "step": 4055
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999713841635386,
+      "loss": 2.639,
+      "step": 4060
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999971313637598,
+      "loss": 2.6309,
+      "step": 4065
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999712430248575,
+      "loss": 2.373,
+      "step": 4070
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999711723253167,
+      "loss": 2.6889,
+      "step": 4075
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999711015389755,
+      "loss": 2.3853,
+      "step": 4080
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999710306658347,
+      "loss": 2.4082,
+      "step": 4085
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999709597058932,
+      "loss": 2.5107,
+      "step": 4090
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999970888659152,
+      "loss": 2.7697,
+      "step": 4095
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999708175256106,
+      "loss": 2.4964,
+      "step": 4100
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999707463052695,
+      "loss": 2.4221,
+      "step": 4105
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999706749981278,
+      "loss": 2.6319,
+      "step": 4110
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999970603604186,
+      "loss": 2.6704,
+      "step": 4115
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999705321234443,
+      "loss": 2.5134,
+      "step": 4120
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999704605559027,
+      "loss": 2.6706,
+      "step": 4125
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999703889015606,
+      "loss": 2.5624,
+      "step": 4130
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999703171604186,
+      "loss": 2.2552,
+      "step": 4135
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999702453324768,
+      "loss": 2.528,
+      "step": 4140
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999970173417735,
+      "loss": 2.7461,
+      "step": 4145
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999701014161928,
+      "loss": 2.5758,
+      "step": 4150
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999700293278508,
+      "loss": 2.63,
+      "step": 4155
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999969957152709,
+      "loss": 2.4362,
+      "step": 4160
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999698848907667,
+      "loss": 2.4211,
+      "step": 4165
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999698125420246,
+      "loss": 2.4353,
+      "step": 4170
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999697401064823,
+      "loss": 2.7979,
+      "step": 4175
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999696675841402,
+      "loss": 2.6134,
+      "step": 4180
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999969594974998,
+      "loss": 2.4935,
+      "step": 4185
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999695222790558,
+      "loss": 2.5087,
+      "step": 4190
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999694494963137,
+      "loss": 2.7233,
+      "step": 4195
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999693766267719,
+      "loss": 2.6594,
+      "step": 4200
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999693036704296,
+      "loss": 2.5277,
+      "step": 4205
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999692306272877,
+      "loss": 2.413,
+      "step": 4210
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999691574973456,
+      "loss": 2.4858,
+      "step": 4215
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999690842806037,
+      "loss": 2.4431,
+      "step": 4220
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999969010977062,
+      "loss": 2.2495,
+      "step": 4225
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000199996893758672,
+      "loss": 2.5721,
+      "step": 4230
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999688641095785,
+      "loss": 2.7159,
+      "step": 4235
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999687905456365,
+      "loss": 2.5314,
+      "step": 4240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999968716894895,
+      "loss": 2.6268,
+      "step": 4245
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999686431573536,
+      "loss": 2.4966,
+      "step": 4250
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999685693330123,
+      "loss": 2.3179,
+      "step": 4255
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999968495421871,
+      "loss": 2.7163,
+      "step": 4260
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999684214239298,
+      "loss": 2.3001,
+      "step": 4265
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999683473391887,
+      "loss": 2.6254,
+      "step": 4270
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999682731676477,
+      "loss": 2.468,
+      "step": 4275
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999681989093067,
+      "loss": 2.5722,
+      "step": 4280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999681245641662,
+      "loss": 2.6148,
+      "step": 4285
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999680501322256,
+      "loss": 2.4819,
+      "step": 4290
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999679756134854,
+      "loss": 2.6615,
+      "step": 4295
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999967901007945,
+      "loss": 2.5589,
+      "step": 4300
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999967826315605,
+      "loss": 2.6975,
+      "step": 4305
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999677515364649,
+      "loss": 2.5358,
+      "step": 4310
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999676766705254,
+      "loss": 2.6952,
+      "step": 4315
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999676017177858,
+      "loss": 2.4479,
+      "step": 4320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999675266782463,
+      "loss": 2.5406,
+      "step": 4325
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999967451551907,
+      "loss": 2.7184,
+      "step": 4330
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999967376338768,
+      "loss": 2.6204,
+      "step": 4335
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999673010388292,
+      "loss": 2.4248,
+      "step": 4340
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999672256520908,
+      "loss": 2.5927,
+      "step": 4345
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999671501785523,
+      "loss": 2.5201,
+      "step": 4350
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999967074618214,
+      "loss": 2.6657,
+      "step": 4355
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999669989710764,
+      "loss": 2.3259,
+      "step": 4360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999669232371385,
+      "loss": 2.7273,
+      "step": 4365
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999966847416401,
+      "loss": 2.7191,
+      "step": 4370
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999966771508864,
+      "loss": 2.5581,
+      "step": 4375
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999666955145272,
+      "loss": 2.721,
+      "step": 4380
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000199996661943339,
+      "loss": 2.5538,
+      "step": 4385
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999665432654537,
+      "loss": 2.5718,
+      "step": 4390
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999966467010718,
+      "loss": 2.435,
+      "step": 4395
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999966390669182,
+      "loss": 2.438,
+      "step": 4400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999663142408465,
+      "loss": 2.6362,
+      "step": 4405
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999966237725711,
+      "loss": 2.6154,
+      "step": 4410
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999661611237762,
+      "loss": 2.343,
+      "step": 4415
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999660844350413,
+      "loss": 2.4809,
+      "step": 4420
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999660076595071,
+      "loss": 2.4247,
+      "step": 4425
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999965930797173,
+      "loss": 2.5994,
+      "step": 4430
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999658538480395,
+      "loss": 2.573,
+      "step": 4435
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999657768121062,
+      "loss": 2.585,
+      "step": 4440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999965699689373,
+      "loss": 2.7138,
+      "step": 4445
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999656224798407,
+      "loss": 2.4242,
+      "step": 4450
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999965545183508,
+      "loss": 2.6523,
+      "step": 4455
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999654678003764,
+      "loss": 2.5392,
+      "step": 4460
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999653903304446,
+      "loss": 2.4328,
+      "step": 4465
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999653127737138,
+      "loss": 2.4089,
+      "step": 4470
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999652351301828,
+      "loss": 2.6982,
+      "step": 4475
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999651573998522,
+      "loss": 2.6046,
+      "step": 4480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999650795827223,
+      "loss": 2.5182,
+      "step": 4485
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999650016787928,
+      "loss": 2.8916,
+      "step": 4490
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999649236880634,
+      "loss": 2.6526,
+      "step": 4495
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999964845610535,
+      "loss": 2.5543,
+      "step": 4500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999647674462064,
+      "loss": 2.5293,
+      "step": 4505
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999646891950785,
+      "loss": 2.4421,
+      "step": 4510
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999964610857151,
+      "loss": 2.8573,
+      "step": 4515
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999964532432424,
+      "loss": 2.5611,
+      "step": 4520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999644539208972,
+      "loss": 2.4044,
+      "step": 4525
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999643753225712,
+      "loss": 2.6072,
+      "step": 4530
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999642966374458,
+      "loss": 2.5995,
+      "step": 4535
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999642178655203,
+      "loss": 2.5983,
+      "step": 4540
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999641390067957,
+      "loss": 2.431,
+      "step": 4545
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999640600612716,
+      "loss": 2.4112,
+      "step": 4550
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999639810289475,
+      "loss": 2.7427,
+      "step": 4555
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999639019098244,
+      "loss": 2.4633,
+      "step": 4560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999638227039017,
+      "loss": 2.7051,
+      "step": 4565
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999637434111795,
+      "loss": 2.6208,
+      "step": 4570
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999636640316578,
+      "loss": 2.3909,
+      "step": 4575
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999635845653366,
+      "loss": 2.6401,
+      "step": 4580
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999635050122158,
+      "loss": 2.544,
+      "step": 4585
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999963425372296,
+      "loss": 2.4398,
+      "step": 4590
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999633456455765,
+      "loss": 2.5029,
+      "step": 4595
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999632658320574,
+      "loss": 2.4968,
+      "step": 4600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999963185931739,
+      "loss": 2.595,
+      "step": 4605
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999963105944621,
+      "loss": 2.5892,
+      "step": 4610
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999630258707036,
+      "loss": 2.418,
+      "step": 4615
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999962945709987,
+      "loss": 2.3198,
+      "step": 4620
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999962865462471,
+      "loss": 2.5866,
+      "step": 4625
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999627851281554,
+      "loss": 2.6689,
+      "step": 4630
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999627047070405,
+      "loss": 2.6745,
+      "step": 4635
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999962624199126,
+      "loss": 2.5815,
+      "step": 4640
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999625436044124,
+      "loss": 2.4527,
+      "step": 4645
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999624629228992,
+      "loss": 2.3316,
+      "step": 4650
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999623821545867,
+      "loss": 2.6772,
+      "step": 4655
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999623012994752,
+      "loss": 2.5055,
+      "step": 4660
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999962220357564,
+      "loss": 2.6098,
+      "step": 4665
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999621393288535,
+      "loss": 2.5823,
+      "step": 4670
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999620582133437,
+      "loss": 2.5924,
+      "step": 4675
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999619770110343,
+      "loss": 2.5865,
+      "step": 4680
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999961895721926,
+      "loss": 2.4947,
+      "step": 4685
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999618143460183,
+      "loss": 2.5983,
+      "step": 4690
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999961732883311,
+      "loss": 2.3177,
+      "step": 4695
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999616513338044,
+      "loss": 2.4433,
+      "step": 4700
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999615696974986,
+      "loss": 2.5015,
+      "step": 4705
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999614879743935,
+      "loss": 2.618,
+      "step": 4710
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999961406164489,
+      "loss": 2.2983,
+      "step": 4715
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999613242677856,
+      "loss": 2.4485,
+      "step": 4720
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999612422842827,
+      "loss": 2.506,
+      "step": 4725
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999611602139808,
+      "loss": 2.7101,
+      "step": 4730
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999961078056879,
+      "loss": 2.6012,
+      "step": 4735
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999609958129785,
+      "loss": 2.6079,
+      "step": 4740
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999609134822787,
+      "loss": 2.6829,
+      "step": 4745
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999608310647794,
+      "loss": 2.6195,
+      "step": 4750
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999607485604812,
+      "loss": 2.3357,
+      "step": 4755
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999606659693836,
+      "loss": 2.6011,
+      "step": 4760
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999960583291487,
+      "loss": 2.7226,
+      "step": 4765
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999960500526791,
+      "loss": 2.5665,
+      "step": 4770
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999604176752957,
+      "loss": 2.5458,
+      "step": 4775
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999603347370013,
+      "loss": 2.6584,
+      "step": 4780
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999602517119076,
+      "loss": 2.8255,
+      "step": 4785
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999601686000151,
+      "loss": 2.5149,
+      "step": 4790
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999960085401323,
+      "loss": 2.5149,
+      "step": 4795
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999960002115832,
+      "loss": 2.529,
+      "step": 4800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999599187435417,
+      "loss": 2.4177,
+      "step": 4805
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999598352844525,
+      "loss": 2.8094,
+      "step": 4810
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999959751738564,
+      "loss": 2.726,
+      "step": 4815
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999959668105876,
+      "loss": 2.493,
+      "step": 4820
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999595843863893,
+      "loss": 2.4468,
+      "step": 4825
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999595005801036,
+      "loss": 2.451,
+      "step": 4830
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999594166870185,
+      "loss": 2.4086,
+      "step": 4835
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999593327071344,
+      "loss": 2.6423,
+      "step": 4840
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999592486404512,
+      "loss": 2.6097,
+      "step": 4845
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999959164486969,
+      "loss": 2.7155,
+      "step": 4850
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999590802466876,
+      "loss": 2.3753,
+      "step": 4855
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999958995919607,
+      "loss": 2.7185,
+      "step": 4860
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999589115057275,
+      "loss": 2.565,
+      "step": 4865
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999588270050488,
+      "loss": 2.4212,
+      "step": 4870
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999587424175715,
+      "loss": 2.5576,
+      "step": 4875
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999586577432944,
+      "loss": 2.5254,
+      "step": 4880
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999958572982219,
+      "loss": 2.3984,
+      "step": 4885
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999958488134344,
+      "loss": 2.5082,
+      "step": 4890
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999584031996702,
+      "loss": 2.6574,
+      "step": 4895
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999583181781977,
+      "loss": 2.3945,
+      "step": 4900
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999582330699255,
+      "loss": 2.5044,
+      "step": 4905
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999958147874855,
+      "loss": 2.6875,
+      "step": 4910
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999958062592985,
+      "loss": 2.5123,
+      "step": 4915
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999579772243163,
+      "loss": 2.4858,
+      "step": 4920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999578917688485,
+      "loss": 2.3913,
+      "step": 4925
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999957806226582,
+      "loss": 2.6742,
+      "step": 4930
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999957720597516,
+      "loss": 2.7184,
+      "step": 4935
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999576348816514,
+      "loss": 2.5892,
+      "step": 4940
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999575490789876,
+      "loss": 2.6998,
+      "step": 4945
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999574631895253,
+      "loss": 2.4694,
+      "step": 4950
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999573772132637,
+      "loss": 2.569,
+      "step": 4955
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999572911502032,
+      "loss": 2.3539,
+      "step": 4960
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999957205000344,
+      "loss": 2.2503,
+      "step": 4965
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999571187636857,
+      "loss": 2.4465,
+      "step": 4970
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999570324402287,
+      "loss": 2.6047,
+      "step": 4975
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999569460299723,
+      "loss": 2.6365,
+      "step": 4980
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999568595329174,
+      "loss": 2.5386,
+      "step": 4985
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999567729490637,
+      "loss": 2.5822,
+      "step": 4990
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999566862784107,
+      "loss": 2.411,
+      "step": 4995
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999956599520959,
+      "loss": 2.9174,
+      "step": 5000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999565126767088,
+      "loss": 2.4363,
+      "step": 5005
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999564257456594,
+      "loss": 2.4856,
+      "step": 5010
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999563387278113,
+      "loss": 2.4739,
+      "step": 5015
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999562516231643,
+      "loss": 2.6323,
+      "step": 5020
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999561644317186,
+      "loss": 2.8467,
+      "step": 5025
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999956077153474,
+      "loss": 2.4187,
+      "step": 5030
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999559897884305,
+      "loss": 2.5242,
+      "step": 5035
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999955902336588,
+      "loss": 2.494,
+      "step": 5040
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999558147979475,
+      "loss": 2.4001,
+      "step": 5045
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999557271725073,
+      "loss": 2.6845,
+      "step": 5050
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999556394602688,
+      "loss": 2.5759,
+      "step": 5055
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999555516612313,
+      "loss": 2.5307,
+      "step": 5060
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999955463775395,
+      "loss": 2.4896,
+      "step": 5065
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999553758027602,
+      "loss": 2.4706,
+      "step": 5070
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999552877433265,
+      "loss": 2.488,
+      "step": 5075
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999955199597094,
+      "loss": 2.5807,
+      "step": 5080
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999955111364063,
+      "loss": 2.6193,
+      "step": 5085
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999955023044233,
+      "loss": 2.4777,
+      "step": 5090
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999549346376045,
+      "loss": 2.662,
+      "step": 5095
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999954846144177,
+      "loss": 2.2384,
+      "step": 5100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999954757563951,
+      "loss": 2.4218,
+      "step": 5105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999954668896926,
+      "loss": 2.507,
+      "step": 5110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999545801431027,
+      "loss": 2.4263,
+      "step": 5115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999544913024806,
+      "loss": 2.6976,
+      "step": 5120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000199995440237506,
+      "loss": 2.6457,
+      "step": 5125
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999543133608402,
+      "loss": 2.4234,
+      "step": 5130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999542242598223,
+      "loss": 2.6219,
+      "step": 5135
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999541350720055,
+      "loss": 2.6571,
+      "step": 5140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000199995404579739,
+      "loss": 2.5993,
+      "step": 5145
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999953956435976,
+      "loss": 2.8249,
+      "step": 5150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999953866987763,
+      "loss": 2.6825,
+      "step": 5155
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999537774527518,
+      "loss": 2.3894,
+      "step": 5160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999536878309416,
+      "loss": 2.356,
+      "step": 5165
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999535981223332,
+      "loss": 2.5617,
+      "step": 5170
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999953508326926,
+      "loss": 2.4124,
+      "step": 5175
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999534184447204,
+      "loss": 2.5813,
+      "step": 5180
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999533284757162,
+      "loss": 2.4507,
+      "step": 5185
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999532384199132,
+      "loss": 2.6176,
+      "step": 5190
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999531482773114,
+      "loss": 2.3542,
+      "step": 5195
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999530580479114,
+      "loss": 2.4879,
+      "step": 5200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999529677317132,
+      "loss": 2.5045,
+      "step": 5205
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999528773287159,
+      "loss": 2.3635,
+      "step": 5210
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999527868389203,
+      "loss": 2.6403,
+      "step": 5215
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999952696262326,
+      "loss": 2.5783,
+      "step": 5220
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999526055989334,
+      "loss": 2.5197,
+      "step": 5225
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999952514848742,
+      "loss": 2.3619,
+      "step": 5230
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999524240117523,
+      "loss": 2.4988,
+      "step": 5235
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999952333087964,
+      "loss": 2.5926,
+      "step": 5240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999522420773773,
+      "loss": 2.2195,
+      "step": 5245
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999952150979992,
+      "loss": 2.4463,
+      "step": 5250
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999520597958084,
+      "loss": 2.4735,
+      "step": 5255
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999519685248262,
+      "loss": 2.4903,
+      "step": 5260
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999518771670455,
+      "loss": 2.4087,
+      "step": 5265
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999517857224666,
+      "loss": 2.7214,
+      "step": 5270
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999951694191089,
+      "loss": 2.8339,
+      "step": 5275
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999516025729134,
+      "loss": 2.3704,
+      "step": 5280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999515108679389,
+      "loss": 2.7017,
+      "step": 5285
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999951419076166,
+      "loss": 2.4457,
+      "step": 5290
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999513271975948,
+      "loss": 2.3882,
+      "step": 5295
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999951235232225,
+      "loss": 2.6735,
+      "step": 5300
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999951143180057,
+      "loss": 2.4836,
+      "step": 5305
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999510510410908,
+      "loss": 2.8034,
+      "step": 5310
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999950958815326,
+      "loss": 2.5311,
+      "step": 5315
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999950866502763,
+      "loss": 2.387,
+      "step": 5320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999507741034014,
+      "loss": 2.3596,
+      "step": 5325
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999506816172416,
+      "loss": 2.6885,
+      "step": 5330
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999505890442833,
+      "loss": 2.6245,
+      "step": 5335
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999504963845268,
+      "loss": 2.5971,
+      "step": 5340
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999950403637972,
+      "loss": 2.6143,
+      "step": 5345
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999503108046187,
+      "loss": 2.2141,
+      "step": 5350
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999950217884467,
+      "loss": 2.3336,
+      "step": 5355
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999501248775173,
+      "loss": 2.3443,
+      "step": 5360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999500317837693,
+      "loss": 2.362,
+      "step": 5365
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999949938603223,
+      "loss": 2.478,
+      "step": 5370
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999498453358782,
+      "loss": 2.6018,
+      "step": 5375
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999497519817354,
+      "loss": 2.7575,
+      "step": 5380
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999496585407938,
+      "loss": 2.6625,
+      "step": 5385
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999495650130546,
+      "loss": 2.6967,
+      "step": 5390
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999949471398517,
+      "loss": 2.3785,
+      "step": 5395
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999949377697181,
+      "loss": 2.5965,
+      "step": 5400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999492839090468,
+      "loss": 2.5036,
+      "step": 5405
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999491900341143,
+      "loss": 2.3819,
+      "step": 5410
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999490960723838,
+      "loss": 2.4559,
+      "step": 5415
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999949002023855,
+      "loss": 2.7719,
+      "step": 5420
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999489078885278,
+      "loss": 2.5455,
+      "step": 5425
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999488136664026,
+      "loss": 2.4121,
+      "step": 5430
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999948719357479,
+      "loss": 2.5166,
+      "step": 5435
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999486249617574,
+      "loss": 2.642,
+      "step": 5440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999485304792377,
+      "loss": 2.3874,
+      "step": 5445
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999484359099195,
+      "loss": 2.492,
+      "step": 5450
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999948341253804,
+      "loss": 2.5342,
+      "step": 5455
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999482465108895,
+      "loss": 2.1746,
+      "step": 5460
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999948151681177,
+      "loss": 2.4772,
+      "step": 5465
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999480567646667,
+      "loss": 2.2773,
+      "step": 5470
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999479617613579,
+      "loss": 2.3527,
+      "step": 5475
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999478666712515,
+      "loss": 2.6607,
+      "step": 5480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999477714943465,
+      "loss": 2.6414,
+      "step": 5485
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999476762306434,
+      "loss": 2.6242,
+      "step": 5490
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999475808801423,
+      "loss": 2.5702,
+      "step": 5495
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999474854428436,
+      "loss": 2.5194,
+      "step": 5500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999473899187463,
+      "loss": 2.3377,
+      "step": 5505
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999947294307851,
+      "loss": 2.2831,
+      "step": 5510
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999947198610158,
+      "loss": 2.5063,
+      "step": 5515
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999471028256665,
+      "loss": 2.6298,
+      "step": 5520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999470069543773,
+      "loss": 2.3577,
+      "step": 5525
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000199994691099629,
+      "loss": 2.4752,
+      "step": 5530
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999468149514045,
+      "loss": 2.7031,
+      "step": 5535
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999467188197212,
+      "loss": 2.514,
+      "step": 5540
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999466226012396,
+      "loss": 2.6673,
+      "step": 5545
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999465262959603,
+      "loss": 2.6299,
+      "step": 5550
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999464299038828,
+      "loss": 2.667,
+      "step": 5555
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999463334250076,
+      "loss": 2.5126,
+      "step": 5560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999946236859334,
+      "loss": 2.5674,
+      "step": 5565
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999946140206863,
+      "loss": 2.278,
+      "step": 5570
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999460434675937,
+      "loss": 2.0909,
+      "step": 5575
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999459466415263,
+      "loss": 2.2677,
+      "step": 5580
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999458497286612,
+      "loss": 2.2392,
+      "step": 5585
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999457527289982,
+      "loss": 2.7081,
+      "step": 5590
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999456556425374,
+      "loss": 2.4179,
+      "step": 5595
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999455584692784,
+      "loss": 2.5304,
+      "step": 5600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999454612092216,
+      "loss": 2.4348,
+      "step": 5605
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999945363862367,
+      "loss": 2.672,
+      "step": 5610
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999452664287143,
+      "loss": 2.226,
+      "step": 5615
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999945168908264,
+      "loss": 2.543,
+      "step": 5620
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999450713010158,
+      "loss": 2.5913,
+      "step": 5625
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999449736069695,
+      "loss": 2.7104,
+      "step": 5630
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999448758261255,
+      "loss": 2.6155,
+      "step": 5635
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999447779584837,
+      "loss": 2.6001,
+      "step": 5640
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999944680004044,
+      "loss": 2.2563,
+      "step": 5645
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999445819628064,
+      "loss": 2.4679,
+      "step": 5650
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999444838347713,
+      "loss": 2.2942,
+      "step": 5655
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999944385619938,
+      "loss": 2.4746,
+      "step": 5660
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999944287318307,
+      "loss": 2.6527,
+      "step": 5665
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999944188929878,
+      "loss": 2.7976,
+      "step": 5670
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999440904546517,
+      "loss": 2.6258,
+      "step": 5675
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999439918926272,
+      "loss": 2.3405,
+      "step": 5680
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999438932438054,
+      "loss": 2.5708,
+      "step": 5685
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999437945081855,
+      "loss": 2.7506,
+      "step": 5690
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999943695685768,
+      "loss": 2.4775,
+      "step": 5695
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999435967765528,
+      "loss": 2.6628,
+      "step": 5700
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999434977805396,
+      "loss": 2.5022,
+      "step": 5705
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999433986977287,
+      "loss": 2.4642,
+      "step": 5710
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999432995281203,
+      "loss": 2.5368,
+      "step": 5715
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999432002717143,
+      "loss": 2.6161,
+      "step": 5720
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999431009285102,
+      "loss": 2.6161,
+      "step": 5725
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999430014985088,
+      "loss": 2.6724,
+      "step": 5730
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999429019817096,
+      "loss": 2.4299,
+      "step": 5735
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999428023781125,
+      "loss": 2.6254,
+      "step": 5740
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999942702687718,
+      "loss": 2.4306,
+      "step": 5745
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999942602910526,
+      "loss": 2.2805,
+      "step": 5750
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999942503046536,
+      "loss": 2.6213,
+      "step": 5755
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999424030957483,
+      "loss": 2.4528,
+      "step": 5760
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999423030581633,
+      "loss": 2.5347,
+      "step": 5765
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999422029337805,
+      "loss": 2.4189,
+      "step": 5770
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999421027226002,
+      "loss": 2.7847,
+      "step": 5775
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999420024246223,
+      "loss": 2.386,
+      "step": 5780
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999419020398467,
+      "loss": 2.3243,
+      "step": 5785
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999418015682733,
+      "loss": 2.501,
+      "step": 5790
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999417010099026,
+      "loss": 2.5756,
+      "step": 5795
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999941600364734,
+      "loss": 2.5002,
+      "step": 5800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999414996327685,
+      "loss": 2.5806,
+      "step": 5805
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999941398814005,
+      "loss": 2.6412,
+      "step": 5810
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999941297908444,
+      "loss": 2.2917,
+      "step": 5815
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999411969160856,
+      "loss": 2.5026,
+      "step": 5820
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999410958369297,
+      "loss": 2.4119,
+      "step": 5825
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999409946709761,
+      "loss": 2.5459,
+      "step": 5830
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999940893418225,
+      "loss": 2.3567,
+      "step": 5835
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999407920786766,
+      "loss": 2.5446,
+      "step": 5840
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999406906523305,
+      "loss": 2.4614,
+      "step": 5845
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999405891391872,
+      "loss": 2.4516,
+      "step": 5850
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999940487539246,
+      "loss": 2.7649,
+      "step": 5855
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999940385852508,
+      "loss": 2.4139,
+      "step": 5860
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999940284078972,
+      "loss": 2.5276,
+      "step": 5865
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999401822186385,
+      "loss": 2.6131,
+      "step": 5870
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999400802715077,
+      "loss": 2.2981,
+      "step": 5875
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999399782375795,
+      "loss": 2.435,
+      "step": 5880
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999398761168542,
+      "loss": 2.4162,
+      "step": 5885
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999397739093312,
+      "loss": 2.4246,
+      "step": 5890
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999396716150107,
+      "loss": 2.6836,
+      "step": 5895
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999939569233893,
+      "loss": 2.537,
+      "step": 5900
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999394667659777,
+      "loss": 2.4223,
+      "step": 5905
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999393642112652,
+      "loss": 2.376,
+      "step": 5910
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999392615697553,
+      "loss": 2.5755,
+      "step": 5915
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999939158841448,
+      "loss": 2.5933,
+      "step": 5920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999390560263434,
+      "loss": 2.0977,
+      "step": 5925
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999389531244414,
+      "loss": 2.705,
+      "step": 5930
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999938850135742,
+      "loss": 2.5654,
+      "step": 5935
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999387470602455,
+      "loss": 2.3034,
+      "step": 5940
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999386438979518,
+      "loss": 2.5059,
+      "step": 5945
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999385406488601,
+      "loss": 2.5865,
+      "step": 5950
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999384373129718,
+      "loss": 2.7423,
+      "step": 5955
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999383338902858,
+      "loss": 2.5133,
+      "step": 5960
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999938230380803,
+      "loss": 2.347,
+      "step": 5965
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999381267845226,
+      "loss": 2.657,
+      "step": 5970
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999380231014452,
+      "loss": 2.4238,
+      "step": 5975
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999379193315703,
+      "loss": 2.4641,
+      "step": 5980
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999378154748982,
+      "loss": 2.5307,
+      "step": 5985
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999937711531429,
+      "loss": 2.5566,
+      "step": 5990
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999376075011623,
+      "loss": 2.5494,
+      "step": 5995
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999375033840985,
+      "loss": 2.4661,
+      "step": 6000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999373991802378,
+      "loss": 2.6792,
+      "step": 6005
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999372948895796,
+      "loss": 2.4679,
+      "step": 6010
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999371905121243,
+      "loss": 2.4224,
+      "step": 6015
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999370860478716,
+      "loss": 2.58,
+      "step": 6020
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999936981496822,
+      "loss": 2.4141,
+      "step": 6025
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999368768589752,
+      "loss": 2.4102,
+      "step": 6030
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999367721343313,
+      "loss": 2.4277,
+      "step": 6035
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999366673228904,
+      "loss": 2.3389,
+      "step": 6040
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999936562424652,
+      "loss": 2.6248,
+      "step": 6045
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999364574396165,
+      "loss": 2.5639,
+      "step": 6050
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999363523677842,
+      "loss": 2.3538,
+      "step": 6055
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999362472091544,
+      "loss": 2.6417,
+      "step": 6060
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999361419637278,
+      "loss": 2.4795,
+      "step": 6065
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999936036631504,
+      "loss": 2.6217,
+      "step": 6070
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999359312124836,
+      "loss": 2.5051,
+      "step": 6075
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999358257066655,
+      "loss": 2.5224,
+      "step": 6080
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999357201140505,
+      "loss": 2.4752,
+      "step": 6085
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999356144346386,
+      "loss": 2.5671,
+      "step": 6090
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999355086684295,
+      "loss": 2.553,
+      "step": 6095
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999354028154233,
+      "loss": 2.7043,
+      "step": 6100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999352968756205,
+      "loss": 2.415,
+      "step": 6105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999351908490202,
+      "loss": 2.5664,
+      "step": 6110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999935084735623,
+      "loss": 2.5261,
+      "step": 6115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999349785354288,
+      "loss": 2.4543,
+      "step": 6120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999348722484378,
+      "loss": 2.5206,
+      "step": 6125
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000199993476587465,
+      "loss": 2.5553,
+      "step": 6130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999346594140648,
+      "loss": 2.4298,
+      "step": 6135
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999934552866683,
+      "loss": 2.5042,
+      "step": 6140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999344462325042,
+      "loss": 2.4607,
+      "step": 6145
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999343395115282,
+      "loss": 2.6575,
+      "step": 6150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999342327037554,
+      "loss": 2.6144,
+      "step": 6155
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999341258091857,
+      "loss": 2.4543,
+      "step": 6160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999340188278192,
+      "loss": 2.6144,
+      "step": 6165
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999339117596557,
+      "loss": 2.8006,
+      "step": 6170
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999933804604695,
+      "loss": 2.7669,
+      "step": 6175
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999336973629378,
+      "loss": 2.722,
+      "step": 6180
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999335900343837,
+      "loss": 2.6892,
+      "step": 6185
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999933482619033,
+      "loss": 2.5959,
+      "step": 6190
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999333751168847,
+      "loss": 2.6445,
+      "step": 6195
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999332675279402,
+      "loss": 2.6802,
+      "step": 6200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999331598521988,
+      "loss": 2.6725,
+      "step": 6205
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999330520896602,
+      "loss": 2.3971,
+      "step": 6210
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999932944240325,
+      "loss": 2.7448,
+      "step": 6215
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999932836304193,
+      "loss": 2.7078,
+      "step": 6220
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999327282812642,
+      "loss": 2.1525,
+      "step": 6225
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999326201715386,
+      "loss": 2.6948,
+      "step": 6230
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999325119750164,
+      "loss": 2.6289,
+      "step": 6235
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999932403691697,
+      "loss": 2.519,
+      "step": 6240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999322953215813,
+      "loss": 2.6539,
+      "step": 6245
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999321868646687,
+      "loss": 2.5327,
+      "step": 6250
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999320783209591,
+      "loss": 2.5113,
+      "step": 6255
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999931969690453,
+      "loss": 2.5714,
+      "step": 6260
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999318609731503,
+      "loss": 2.4982,
+      "step": 6265
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999931752169051,
+      "loss": 2.4783,
+      "step": 6270
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999316432781544,
+      "loss": 2.4137,
+      "step": 6275
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999315343004616,
+      "loss": 2.3633,
+      "step": 6280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999931425235972,
+      "loss": 2.4293,
+      "step": 6285
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999313160846854,
+      "loss": 2.4783,
+      "step": 6290
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999312068466024,
+      "loss": 2.4389,
+      "step": 6295
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999310975217228,
+      "loss": 2.4359,
+      "step": 6300
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999930988110047,
+      "loss": 2.7623,
+      "step": 6305
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999308786115738,
+      "loss": 2.5565,
+      "step": 6310
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999930769026304,
+      "loss": 2.5909,
+      "step": 6315
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999930659354238,
+      "loss": 2.4685,
+      "step": 6320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999305495953753,
+      "loss": 2.3457,
+      "step": 6325
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999930439749716,
+      "loss": 2.4567,
+      "step": 6330
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999303298172597,
+      "loss": 2.6181,
+      "step": 6335
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999302197980075,
+      "loss": 2.6353,
+      "step": 6340
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999301096919583,
+      "loss": 2.7282,
+      "step": 6345
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999299994991128,
+      "loss": 2.2838,
+      "step": 6350
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999298892194704,
+      "loss": 2.492,
+      "step": 6355
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999929778853032,
+      "loss": 2.7185,
+      "step": 6360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999296683997962,
+      "loss": 2.4134,
+      "step": 6365
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999295578597645,
+      "loss": 2.3234,
+      "step": 6370
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999294472329362,
+      "loss": 2.2621,
+      "step": 6375
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999293365193115,
+      "loss": 2.5966,
+      "step": 6380
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999292257188902,
+      "loss": 2.3967,
+      "step": 6385
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999929114831672,
+      "loss": 2.5198,
+      "step": 6390
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999929003857658,
+      "loss": 2.7163,
+      "step": 6395
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999928892796847,
+      "loss": 2.6591,
+      "step": 6400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000199992878164924,
+      "loss": 2.4431,
+      "step": 6405
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999286704148363,
+      "loss": 2.5429,
+      "step": 6410
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999928559093636,
+      "loss": 2.476,
+      "step": 6415
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999284476856396,
+      "loss": 2.7277,
+      "step": 6420
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999283361908467,
+      "loss": 2.5014,
+      "step": 6425
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999282246092575,
+      "loss": 2.3429,
+      "step": 6430
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999281129408716,
+      "loss": 2.4629,
+      "step": 6435
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999280011856897,
+      "loss": 2.5457,
+      "step": 6440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999927889343711,
+      "loss": 2.72,
+      "step": 6445
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999277774149362,
+      "loss": 2.6957,
+      "step": 6450
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999927665399365,
+      "loss": 2.7006,
+      "step": 6455
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999275532969974,
+      "loss": 2.1886,
+      "step": 6460
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999274411078335,
+      "loss": 2.6278,
+      "step": 6465
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999273288318734,
+      "loss": 2.423,
+      "step": 6470
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999272164691168,
+      "loss": 2.4197,
+      "step": 6475
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999271040195638,
+      "loss": 2.4111,
+      "step": 6480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999269914832145,
+      "loss": 2.4731,
+      "step": 6485
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999926878860069,
+      "loss": 2.7607,
+      "step": 6490
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999267661501274,
+      "loss": 2.5821,
+      "step": 6495
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999266533533893,
+      "loss": 2.6406,
+      "step": 6500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999926540469855,
+      "loss": 2.4918,
+      "step": 6505
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999264274995246,
+      "loss": 2.5968,
+      "step": 6510
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999926314442398,
+      "loss": 2.6865,
+      "step": 6515
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999262012984748,
+      "loss": 2.4477,
+      "step": 6520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999260880677557,
+      "loss": 2.7181,
+      "step": 6525
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000199992597475024,
+      "loss": 2.6592,
+      "step": 6530
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999258613459287,
+      "loss": 2.4172,
+      "step": 6535
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999257478548206,
+      "loss": 2.5165,
+      "step": 6540
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999256342769165,
+      "loss": 2.2554,
+      "step": 6545
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999255206122165,
+      "loss": 2.2255,
+      "step": 6550
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000199992540686072,
+      "loss": 2.6928,
+      "step": 6555
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999252930224276,
+      "loss": 2.5198,
+      "step": 6560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999925179097339,
+      "loss": 2.4785,
+      "step": 6565
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999925065085454,
+      "loss": 2.444,
+      "step": 6570
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999249509867735,
+      "loss": 2.2624,
+      "step": 6575
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999248368012963,
+      "loss": 2.5451,
+      "step": 6580
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999247225290233,
+      "loss": 2.5046,
+      "step": 6585
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999246081699542,
+      "loss": 2.651,
+      "step": 6590
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999244937240888,
+      "loss": 2.5855,
+      "step": 6595
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999243791914273,
+      "loss": 2.6744,
+      "step": 6600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999242645719697,
+      "loss": 2.4147,
+      "step": 6605
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999241498657163,
+      "loss": 2.4975,
+      "step": 6610
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999240350726668,
+      "loss": 2.5738,
+      "step": 6615
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999239201928215,
+      "loss": 2.6384,
+      "step": 6620
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000199992380522618,
+      "loss": 2.4977,
+      "step": 6625
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999923690172742,
+      "loss": 2.4549,
+      "step": 6630
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999235750325087,
+      "loss": 2.5053,
+      "step": 6635
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999234598054791,
+      "loss": 2.6512,
+      "step": 6640
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999233444916535,
+      "loss": 2.6486,
+      "step": 6645
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999923229091032,
+      "loss": 2.6118,
+      "step": 6650
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999231136036145,
+      "loss": 2.6748,
+      "step": 6655
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999922998029401,
+      "loss": 2.4618,
+      "step": 6660
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999228823683915,
+      "loss": 2.667,
+      "step": 6665
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999227666205862,
+      "loss": 2.5785,
+      "step": 6670
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999922650785985,
+      "loss": 2.3421,
+      "step": 6675
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999225348645876,
+      "loss": 2.5338,
+      "step": 6680
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999224188563947,
+      "loss": 2.378,
+      "step": 6685
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999223027614057,
+      "loss": 2.5207,
+      "step": 6690
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999922186579621,
+      "loss": 2.4966,
+      "step": 6695
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000199992207031104,
+      "loss": 2.6312,
+      "step": 6700
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999219539556636,
+      "loss": 2.7636,
+      "step": 6705
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999218375134912,
+      "loss": 2.5626,
+      "step": 6710
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999217209845226,
+      "loss": 2.6757,
+      "step": 6715
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999216043687585,
+      "loss": 2.3545,
+      "step": 6720
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999214876661986,
+      "loss": 2.4726,
+      "step": 6725
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999213708768427,
+      "loss": 2.7951,
+      "step": 6730
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999212540006912,
+      "loss": 2.4741,
+      "step": 6735
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999921137037744,
+      "loss": 2.2713,
+      "step": 6740
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999210199880008,
+      "loss": 2.3387,
+      "step": 6745
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999920902851462,
+      "loss": 2.4719,
+      "step": 6750
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999207856281271,
+      "loss": 2.5354,
+      "step": 6755
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999920668317997,
+      "loss": 2.3241,
+      "step": 6760
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999205509210706,
+      "loss": 2.6475,
+      "step": 6765
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999204334373487,
+      "loss": 2.554,
+      "step": 6770
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999920315866831,
+      "loss": 2.4501,
+      "step": 6775
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999201982095178,
+      "loss": 2.3314,
+      "step": 6780
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999200804654088,
+      "loss": 2.4591,
+      "step": 6785
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999919962634504,
+      "loss": 2.603,
+      "step": 6790
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999198447168034,
+      "loss": 2.4301,
+      "step": 6795
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999197267123075,
+      "loss": 2.4142,
+      "step": 6800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999196086210158,
+      "loss": 2.4162,
+      "step": 6805
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999194904429284,
+      "loss": 2.4825,
+      "step": 6810
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999193721780456,
+      "loss": 2.3559,
+      "step": 6815
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999192538263668,
+      "loss": 2.3979,
+      "step": 6820
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999191353878924,
+      "loss": 2.355,
+      "step": 6825
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999190168626228,
+      "loss": 2.3177,
+      "step": 6830
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999918898250557,
+      "loss": 2.7116,
+      "step": 6835
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999187795516961,
+      "loss": 2.7235,
+      "step": 6840
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999186607660394,
+      "loss": 2.5057,
+      "step": 6845
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999185418935873,
+      "loss": 2.5062,
+      "step": 6850
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999184229343395,
+      "loss": 2.5559,
+      "step": 6855
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999183038882961,
+      "loss": 2.4072,
+      "step": 6860
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999181847554572,
+      "loss": 2.3689,
+      "step": 6865
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999918065535823,
+      "loss": 2.3674,
+      "step": 6870
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999917946229393,
+      "loss": 2.4241,
+      "step": 6875
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999917826836168,
+      "loss": 2.6018,
+      "step": 6880
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999917707356147,
+      "loss": 2.5347,
+      "step": 6885
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999175877893305,
+      "loss": 2.3115,
+      "step": 6890
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999174681357187,
+      "loss": 2.5077,
+      "step": 6895
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999173483953113,
+      "loss": 2.6317,
+      "step": 6900
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999172285681084,
+      "loss": 2.5362,
+      "step": 6905
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999171086541105,
+      "loss": 2.5657,
+      "step": 6910
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999169886533168,
+      "loss": 2.5338,
+      "step": 6915
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999168685657276,
+      "loss": 2.6462,
+      "step": 6920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999916748391343,
+      "loss": 2.5496,
+      "step": 6925
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999166281301634,
+      "loss": 2.629,
+      "step": 6930
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999165077821883,
+      "loss": 2.7223,
+      "step": 6935
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999163873474175,
+      "loss": 2.3582,
+      "step": 6940
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999162668258514,
+      "loss": 2.4163,
+      "step": 6945
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000199991614621749,
+      "loss": 2.6175,
+      "step": 6950
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999160255223335,
+      "loss": 2.5896,
+      "step": 6955
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999159047403816,
+      "loss": 2.5343,
+      "step": 6960
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999157838716342,
+      "loss": 2.6274,
+      "step": 6965
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999156629160912,
+      "loss": 2.578,
+      "step": 6970
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999155418737533,
+      "loss": 2.5981,
+      "step": 6975
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999154207446198,
+      "loss": 2.4209,
+      "step": 6980
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999152995286914,
+      "loss": 2.5243,
+      "step": 6985
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999151782259676,
+      "loss": 2.7483,
+      "step": 6990
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999915056836448,
+      "loss": 2.4876,
+      "step": 6995
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999149353601338,
+      "loss": 2.6989,
+      "step": 7000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999148137970243,
+      "loss": 2.634,
+      "step": 7005
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999146921471196,
+      "loss": 2.4256,
+      "step": 7010
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999145704104195,
+      "loss": 2.4754,
+      "step": 7015
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999144485869242,
+      "loss": 2.4822,
+      "step": 7020
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999143266766336,
+      "loss": 2.5492,
+      "step": 7025
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999914204679548,
+      "loss": 2.5732,
+      "step": 7030
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999914082595667,
+      "loss": 2.2351,
+      "step": 7035
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999913960424991,
+      "loss": 2.6789,
+      "step": 7040
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999138381675198,
+      "loss": 2.5905,
+      "step": 7045
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999137158232535,
+      "loss": 2.6706,
+      "step": 7050
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999913593392192,
+      "loss": 2.4164,
+      "step": 7055
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999134708743353,
+      "loss": 2.6302,
+      "step": 7060
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999133482696837,
+      "loss": 2.5643,
+      "step": 7065
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999913225578237,
+      "loss": 2.6243,
+      "step": 7070
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999131027999948,
+      "loss": 2.4086,
+      "step": 7075
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999912979934958,
+      "loss": 2.6565,
+      "step": 7080
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999128569831259,
+      "loss": 2.6168,
+      "step": 7085
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999127339444988,
+      "loss": 2.4882,
+      "step": 7090
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999126108190764,
+      "loss": 2.2434,
+      "step": 7095
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999912487606859,
+      "loss": 2.5088,
+      "step": 7100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999123643078468,
+      "loss": 2.662,
+      "step": 7105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999122409220394,
+      "loss": 2.3635,
+      "step": 7110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999121174494374,
+      "loss": 2.5612,
+      "step": 7115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999119938900398,
+      "loss": 2.4165,
+      "step": 7120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999118702438478,
+      "loss": 2.5879,
+      "step": 7125
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999117465108602,
+      "loss": 2.5018,
+      "step": 7130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999116226910782,
+      "loss": 2.6596,
+      "step": 7135
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999911498784501,
+      "loss": 2.6165,
+      "step": 7140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999911374791129,
+      "loss": 2.2687,
+      "step": 7145
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999911250710962,
+      "loss": 2.7033,
+      "step": 7150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999111265439997,
+      "loss": 2.4913,
+      "step": 7155
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999110022902428,
+      "loss": 2.7355,
+      "step": 7160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999910877949691,
+      "loss": 2.4507,
+      "step": 7165
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999107535223444,
+      "loss": 2.5272,
+      "step": 7170
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999106290082027,
+      "loss": 2.749,
+      "step": 7175
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999910504407266,
+      "loss": 2.4736,
+      "step": 7180
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999910379719535,
+      "loss": 2.7673,
+      "step": 7185
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999102549450086,
+      "loss": 2.6617,
+      "step": 7190
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999101300836877,
+      "loss": 2.303,
+      "step": 7195
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999910005135572,
+      "loss": 3.012,
+      "step": 7200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999098801006614,
+      "loss": 2.5122,
+      "step": 7205
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999097549789558,
+      "loss": 2.4943,
+      "step": 7210
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999096297704556,
+      "loss": 2.3182,
+      "step": 7215
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999095044751606,
+      "loss": 2.5661,
+      "step": 7220
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999093790930706,
+      "loss": 2.3979,
+      "step": 7225
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999092536241862,
+      "loss": 2.4857,
+      "step": 7230
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999091280685068,
+      "loss": 2.2486,
+      "step": 7235
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999909002426033,
+      "loss": 2.5288,
+      "step": 7240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999908876696764,
+      "loss": 2.5035,
+      "step": 7245
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999087508807003,
+      "loss": 2.4987,
+      "step": 7250
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999086249778425,
+      "loss": 2.4817,
+      "step": 7255
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999084989881895,
+      "loss": 2.4187,
+      "step": 7260
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999083729117418,
+      "loss": 2.4947,
+      "step": 7265
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999082467484996,
+      "loss": 2.5637,
+      "step": 7270
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999081204984627,
+      "loss": 2.6972,
+      "step": 7275
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999907994161631,
+      "loss": 2.4978,
+      "step": 7280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999907867738005,
+      "loss": 2.61,
+      "step": 7285
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999077412275842,
+      "loss": 2.674,
+      "step": 7290
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999076146303685,
+      "loss": 2.8544,
+      "step": 7295
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999074879463585,
+      "loss": 2.3708,
+      "step": 7300
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999073611755538,
+      "loss": 2.5549,
+      "step": 7305
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999072343179547,
+      "loss": 2.4773,
+      "step": 7310
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999071073735606,
+      "loss": 2.4491,
+      "step": 7315
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999069803423725,
+      "loss": 2.3739,
+      "step": 7320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999068532243895,
+      "loss": 2.5656,
+      "step": 7325
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999067260196118,
+      "loss": 2.2732,
+      "step": 7330
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999065987280398,
+      "loss": 2.3354,
+      "step": 7335
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999064713496735,
+      "loss": 2.5572,
+      "step": 7340
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999063438845124,
+      "loss": 2.4688,
+      "step": 7345
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999062163325565,
+      "loss": 2.4249,
+      "step": 7350
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999060886938068,
+      "loss": 2.4774,
+      "step": 7355
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999905960968262,
+      "loss": 2.523,
+      "step": 7360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999905833155923,
+      "loss": 2.5578,
+      "step": 7365
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000199990570525679,
+      "loss": 2.3641,
+      "step": 7370
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999055772708618,
+      "loss": 2.4622,
+      "step": 7375
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999054491981398,
+      "loss": 2.6698,
+      "step": 7380
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999905321038623,
+      "loss": 2.5802,
+      "step": 7385
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999051927923115,
+      "loss": 2.5884,
+      "step": 7390
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999050644592062,
+      "loss": 2.4902,
+      "step": 7395
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999049360393064,
+      "loss": 2.7938,
+      "step": 7400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999904807532612,
+      "loss": 2.5248,
+      "step": 7405
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999046789391232,
+      "loss": 2.6136,
+      "step": 7410
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999045502588404,
+      "loss": 2.627,
+      "step": 7415
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999044214917628,
+      "loss": 2.4668,
+      "step": 7420
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999042926378913,
+      "loss": 2.3442,
+      "step": 7425
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999041636972251,
+      "loss": 2.4571,
+      "step": 7430
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999040346697648,
+      "loss": 2.5193,
+      "step": 7435
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999039055555105,
+      "loss": 2.8178,
+      "step": 7440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999037763544615,
+      "loss": 2.5323,
+      "step": 7445
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999036470666183,
+      "loss": 2.3081,
+      "step": 7450
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999035176919807,
+      "loss": 2.4393,
+      "step": 7455
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999033882305491,
+      "loss": 2.5866,
+      "step": 7460
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999032586823232,
+      "loss": 2.6877,
+      "step": 7465
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999903129047303,
+      "loss": 2.3796,
+      "step": 7470
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999029993254886,
+      "loss": 2.5524,
+      "step": 7475
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000199990286951688,
+      "loss": 2.5185,
+      "step": 7480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999027396214771,
+      "loss": 2.742,
+      "step": 7485
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000199990260963928,
+      "loss": 2.492,
+      "step": 7490
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999902479570289,
+      "loss": 2.6437,
+      "step": 7495
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999023494145037,
+      "loss": 2.5175,
+      "step": 7500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999022191719243,
+      "loss": 2.0803,
+      "step": 7505
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999020888425507,
+      "loss": 2.5532,
+      "step": 7510
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999901958426383,
+      "loss": 2.4239,
+      "step": 7515
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999018279234212,
+      "loss": 2.2999,
+      "step": 7520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999901697333665,
+      "loss": 2.3891,
+      "step": 7525
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999901566657115,
+      "loss": 2.6701,
+      "step": 7530
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999014358937708,
+      "loss": 2.6404,
+      "step": 7535
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999013050436327,
+      "loss": 2.4318,
+      "step": 7540
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999011741067007,
+      "loss": 2.5055,
+      "step": 7545
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999010430829742,
+      "loss": 2.5732,
+      "step": 7550
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999009119724538,
+      "loss": 2.5331,
+      "step": 7555
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999007807751395,
+      "loss": 2.3426,
+      "step": 7560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999900649491031,
+      "loss": 2.5285,
+      "step": 7565
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999005181201287,
+      "loss": 2.2141,
+      "step": 7570
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999003866624324,
+      "loss": 2.3738,
+      "step": 7575
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999900255117942,
+      "loss": 2.138,
+      "step": 7580
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999001234866578,
+      "loss": 2.6575,
+      "step": 7585
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998999917685795,
+      "loss": 2.4234,
+      "step": 7590
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999899859963707,
+      "loss": 2.4141,
+      "step": 7595
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999899728072041,
+      "loss": 2.683,
+      "step": 7600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999899596093581,
+      "loss": 2.6398,
+      "step": 7605
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998994640283267,
+      "loss": 2.4464,
+      "step": 7610
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999899331876279,
+      "loss": 2.5567,
+      "step": 7615
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999899199637437,
+      "loss": 2.4854,
+      "step": 7620
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998990673118016,
+      "loss": 2.6837,
+      "step": 7625
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998989348993717,
+      "loss": 2.4272,
+      "step": 7630
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998988024001486,
+      "loss": 2.5749,
+      "step": 7635
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998986698141312,
+      "loss": 2.38,
+      "step": 7640
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998985371413202,
+      "loss": 2.5107,
+      "step": 7645
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998984043817153,
+      "loss": 2.5564,
+      "step": 7650
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998982715353165,
+      "loss": 2.4367,
+      "step": 7655
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999898138602124,
+      "loss": 2.4698,
+      "step": 7660
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998980055821377,
+      "loss": 2.5996,
+      "step": 7665
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998978724753577,
+      "loss": 2.4685,
+      "step": 7670
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998977392817838,
+      "loss": 2.4938,
+      "step": 7675
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998976060014162,
+      "loss": 2.3933,
+      "step": 7680
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999897472634255,
+      "loss": 2.3589,
+      "step": 7685
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998973391802997,
+      "loss": 2.526,
+      "step": 7690
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999897205639551,
+      "loss": 2.3884,
+      "step": 7695
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998970720120087,
+      "loss": 2.6155,
+      "step": 7700
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998969382976727,
+      "loss": 2.347,
+      "step": 7705
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998968044965428,
+      "loss": 2.3685,
+      "step": 7710
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999896670608619,
+      "loss": 2.6701,
+      "step": 7715
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999896536633902,
+      "loss": 2.5651,
+      "step": 7720
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998964025723913,
+      "loss": 2.323,
+      "step": 7725
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998962684240868,
+      "loss": 2.6822,
+      "step": 7730
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998961341889885,
+      "loss": 2.4658,
+      "step": 7735
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999895999867097,
+      "loss": 2.5973,
+      "step": 7740
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998958654584117,
+      "loss": 2.3331,
+      "step": 7745
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999895730962933,
+      "loss": 2.4025,
+      "step": 7750
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998955963806606,
+      "loss": 2.4536,
+      "step": 7755
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998954617115943,
+      "loss": 2.6764,
+      "step": 7760
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998953269557347,
+      "loss": 2.4254,
+      "step": 7765
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998951921130814,
+      "loss": 2.4398,
+      "step": 7770
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999895057183635,
+      "loss": 2.5879,
+      "step": 7775
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998949221673948,
+      "loss": 2.1881,
+      "step": 7780
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998947870643612,
+      "loss": 2.5155,
+      "step": 7785
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998946518745342,
+      "loss": 2.3662,
+      "step": 7790
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998945165979133,
+      "loss": 2.3525,
+      "step": 7795
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998943812344993,
+      "loss": 2.5677,
+      "step": 7800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998942457842916,
+      "loss": 2.4675,
+      "step": 7805
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999894110247291,
+      "loss": 2.5602,
+      "step": 7810
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998939746234963,
+      "loss": 2.5901,
+      "step": 7815
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998938389129083,
+      "loss": 2.701,
+      "step": 7820
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999893703115527,
+      "loss": 2.6018,
+      "step": 7825
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998935672313524,
+      "loss": 2.4727,
+      "step": 7830
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999893431260384,
+      "loss": 2.5026,
+      "step": 7835
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998932952026229,
+      "loss": 2.5641,
+      "step": 7840
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998931590580678,
+      "loss": 2.6295,
+      "step": 7845
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998930228267198,
+      "loss": 2.4993,
+      "step": 7850
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998928865085783,
+      "loss": 2.7407,
+      "step": 7855
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998927501036433,
+      "loss": 2.6448,
+      "step": 7860
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999892613611915,
+      "loss": 2.5057,
+      "step": 7865
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998924770333937,
+      "loss": 2.4465,
+      "step": 7870
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999892340368079,
+      "loss": 2.6736,
+      "step": 7875
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998922036159708,
+      "loss": 2.4964,
+      "step": 7880
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998920667770696,
+      "loss": 2.4419,
+      "step": 7885
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999891929851375,
+      "loss": 2.6466,
+      "step": 7890
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999891792838887,
+      "loss": 2.3967,
+      "step": 7895
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999891655739606,
+      "loss": 2.5252,
+      "step": 7900
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999891518553532,
+      "loss": 2.5781,
+      "step": 7905
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998913812806644,
+      "loss": 2.2417,
+      "step": 7910
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998912439210035,
+      "loss": 2.3379,
+      "step": 7915
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998911064745496,
+      "loss": 2.6051,
+      "step": 7920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998909689413025,
+      "loss": 2.498,
+      "step": 7925
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998908313212624,
+      "loss": 2.4427,
+      "step": 7930
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999890693614429,
+      "loss": 2.3317,
+      "step": 7935
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998905558208025,
+      "loss": 2.4034,
+      "step": 7940
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999890417940383,
+      "loss": 2.5472,
+      "step": 7945
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998902799731703,
+      "loss": 2.4709,
+      "step": 7950
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999890141919164,
+      "loss": 2.5028,
+      "step": 7955
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998900037783654,
+      "loss": 2.2752,
+      "step": 7960
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998898655507733,
+      "loss": 2.2431,
+      "step": 7965
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998897272363882,
+      "loss": 2.5253,
+      "step": 7970
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998895888352102,
+      "loss": 2.6414,
+      "step": 7975
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998894503472388,
+      "loss": 2.7794,
+      "step": 7980
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998893117724746,
+      "loss": 2.5362,
+      "step": 7985
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998891731109173,
+      "loss": 2.4186,
+      "step": 7990
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998890343625672,
+      "loss": 2.6099,
+      "step": 7995
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998888955274238,
+      "loss": 2.505,
+      "step": 8000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998887566054875,
+      "loss": 2.499,
+      "step": 8005
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998886175967584,
+      "loss": 2.7635,
+      "step": 8010
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998884785012362,
+      "loss": 2.4564,
+      "step": 8015
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998883393189208,
+      "loss": 2.5485,
+      "step": 8020
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998882000498133,
+      "loss": 2.5103,
+      "step": 8025
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999888060693912,
+      "loss": 2.5329,
+      "step": 8030
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998879212512183,
+      "loss": 2.6389,
+      "step": 8035
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998877817217314,
+      "loss": 2.4754,
+      "step": 8040
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998876421054517,
+      "loss": 2.5635,
+      "step": 8045
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998875024023792,
+      "loss": 2.3957,
+      "step": 8050
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998873626125136,
+      "loss": 2.5177,
+      "step": 8055
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998872227358554,
+      "loss": 2.5145,
+      "step": 8060
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998870827724044,
+      "loss": 2.6447,
+      "step": 8065
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998869427221604,
+      "loss": 2.4512,
+      "step": 8070
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998868025851237,
+      "loss": 2.4837,
+      "step": 8075
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998866623612943,
+      "loss": 2.4513,
+      "step": 8080
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999886522050672,
+      "loss": 2.3954,
+      "step": 8085
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998863816532566,
+      "loss": 2.4231,
+      "step": 8090
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999886241169049,
+      "loss": 2.4992,
+      "step": 8095
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998861005980482,
+      "loss": 2.4991,
+      "step": 8100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999885959940255,
+      "loss": 2.3801,
+      "step": 8105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998858191956688,
+      "loss": 2.6345,
+      "step": 8110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000199988567836429,
+      "loss": 2.7197,
+      "step": 8115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998855374461186,
+      "loss": 2.3708,
+      "step": 8120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998853964411543,
+      "loss": 2.0928,
+      "step": 8125
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998852553493974,
+      "loss": 2.3053,
+      "step": 8130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999885114170848,
+      "loss": 2.6692,
+      "step": 8135
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998849729055057,
+      "loss": 2.6266,
+      "step": 8140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999884831553371,
+      "loss": 2.5056,
+      "step": 8145
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998846901144433,
+      "loss": 2.2743,
+      "step": 8150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999884548588723,
+      "loss": 2.4226,
+      "step": 8155
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998844069762104,
+      "loss": 2.6256,
+      "step": 8160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999884265276905,
+      "loss": 2.4524,
+      "step": 8165
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998841234908073,
+      "loss": 2.5078,
+      "step": 8170
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999883981617917,
+      "loss": 2.5066,
+      "step": 8175
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998838396582337,
+      "loss": 2.4922,
+      "step": 8180
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998836976117584,
+      "loss": 2.3698,
+      "step": 8185
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998835554784904,
+      "loss": 2.5426,
+      "step": 8190
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998834132584295,
+      "loss": 2.4569,
+      "step": 8195
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998832709515764,
+      "loss": 2.5148,
+      "step": 8200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999883128557931,
+      "loss": 2.4442,
+      "step": 8205
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998829860774927,
+      "loss": 2.4735,
+      "step": 8210
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998828435102624,
+      "loss": 2.6126,
+      "step": 8215
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998827008562393,
+      "loss": 2.5147,
+      "step": 8220
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998825581154237,
+      "loss": 2.354,
+      "step": 8225
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999882415287816,
+      "loss": 2.4763,
+      "step": 8230
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998822723734156,
+      "loss": 2.4824,
+      "step": 8235
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998821293722228,
+      "loss": 2.6852,
+      "step": 8240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998819862842378,
+      "loss": 2.4895,
+      "step": 8245
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998818431094605,
+      "loss": 2.4536,
+      "step": 8250
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998816998478903,
+      "loss": 2.7339,
+      "step": 8255
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998815564995282,
+      "loss": 2.6485,
+      "step": 8260
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998814130643735,
+      "loss": 2.599,
+      "step": 8265
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998812695424265,
+      "loss": 2.4026,
+      "step": 8270
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998811259336875,
+      "loss": 2.5381,
+      "step": 8275
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999880982238156,
+      "loss": 2.6003,
+      "step": 8280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999880838455832,
+      "loss": 2.2663,
+      "step": 8285
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999880694586716,
+      "loss": 2.6823,
+      "step": 8290
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998805506308073,
+      "loss": 2.3228,
+      "step": 8295
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999880406588107,
+      "loss": 2.3697,
+      "step": 8300
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999880262458614,
+      "loss": 2.347,
+      "step": 8305
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998801182423287,
+      "loss": 2.5555,
+      "step": 8310
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998799739392515,
+      "loss": 2.4845,
+      "step": 8315
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998798295493817,
+      "loss": 2.6127,
+      "step": 8320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000199987968507272,
+      "loss": 2.5411,
+      "step": 8325
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999879540509266,
+      "loss": 2.4625,
+      "step": 8330
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000199987939585902,
+      "loss": 2.2127,
+      "step": 8335
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998792511219816,
+      "loss": 2.3251,
+      "step": 8340
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998791062981512,
+      "loss": 2.4014,
+      "step": 8345
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998789613875285,
+      "loss": 2.4815,
+      "step": 8350
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999878816390114,
+      "loss": 2.2377,
+      "step": 8355
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998786713059071,
+      "loss": 2.4975,
+      "step": 8360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998785261349084,
+      "loss": 2.3934,
+      "step": 8365
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998783808771172,
+      "loss": 2.2827,
+      "step": 8370
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998782355325342,
+      "loss": 2.282,
+      "step": 8375
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998780901011595,
+      "loss": 2.7165,
+      "step": 8380
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998779445829922,
+      "loss": 2.3389,
+      "step": 8385
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999877798978033,
+      "loss": 2.4213,
+      "step": 8390
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998776532862818,
+      "loss": 2.7093,
+      "step": 8395
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998775075077385,
+      "loss": 2.3988,
+      "step": 8400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998773616424035,
+      "loss": 2.4379,
+      "step": 8405
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999877215690276,
+      "loss": 2.382,
+      "step": 8410
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998770696513568,
+      "loss": 2.3973,
+      "step": 8415
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998769235256457,
+      "loss": 2.5994,
+      "step": 8420
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999876777313143,
+      "loss": 2.3849,
+      "step": 8425
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999876631013848,
+      "loss": 2.4507,
+      "step": 8430
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998764846277612,
+      "loss": 2.3732,
+      "step": 8435
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999876338154882,
+      "loss": 2.2734,
+      "step": 8440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998761915952112,
+      "loss": 2.6901,
+      "step": 8445
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999876044948749,
+      "loss": 2.5284,
+      "step": 8450
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998758982154942,
+      "loss": 2.4051,
+      "step": 8455
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999875751395448,
+      "loss": 2.6417,
+      "step": 8460
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000199987560448861,
+      "loss": 2.4962,
+      "step": 8465
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000199987545749498,
+      "loss": 2.5253,
+      "step": 8470
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999875310414558,
+      "loss": 2.3919,
+      "step": 8475
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998751632473445,
+      "loss": 2.3322,
+      "step": 8480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998750159933392,
+      "loss": 2.4986,
+      "step": 8485
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998748686525422,
+      "loss": 2.4157,
+      "step": 8490
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999874721224953,
+      "loss": 2.2829,
+      "step": 8495
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998745737105723,
+      "loss": 2.4797,
+      "step": 8500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998744261094,
+      "loss": 2.3625,
+      "step": 8505
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999874278421436,
+      "loss": 2.4289,
+      "step": 8510
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000199987413064668,
+      "loss": 2.5305,
+      "step": 8515
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998739827851325,
+      "loss": 2.3624,
+      "step": 8520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998738348367932,
+      "loss": 2.4507,
+      "step": 8525
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998736868016622,
+      "loss": 2.438,
+      "step": 8530
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998735386797395,
+      "loss": 2.5493,
+      "step": 8535
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998733904710254,
+      "loss": 2.6027,
+      "step": 8540
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998732421755194,
+      "loss": 2.6591,
+      "step": 8545
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999873093793222,
+      "loss": 2.5961,
+      "step": 8550
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998729453241326,
+      "loss": 2.6149,
+      "step": 8555
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998727967682523,
+      "loss": 2.4328,
+      "step": 8560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998726481255797,
+      "loss": 2.7167,
+      "step": 8565
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999872499396116,
+      "loss": 2.621,
+      "step": 8570
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998723505798604,
+      "loss": 2.4871,
+      "step": 8575
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998722016768134,
+      "loss": 2.4709,
+      "step": 8580
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999872052686975,
+      "loss": 2.639,
+      "step": 8585
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999871903610345,
+      "loss": 2.4479,
+      "step": 8590
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998717544469235,
+      "loss": 2.4216,
+      "step": 8595
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998716051967104,
+      "loss": 2.7285,
+      "step": 8600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998714558597057,
+      "loss": 2.6347,
+      "step": 8605
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000199987130643591,
+      "loss": 2.7399,
+      "step": 8610
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998711569253224,
+      "loss": 2.4631,
+      "step": 8615
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998710073279433,
+      "loss": 2.6147,
+      "step": 8620
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999870857643773,
+      "loss": 2.7385,
+      "step": 8625
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998707078728111,
+      "loss": 2.5324,
+      "step": 8630
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998705580150583,
+      "loss": 2.6185,
+      "step": 8635
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998704080705136,
+      "loss": 2.4729,
+      "step": 8640
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998702580391778,
+      "loss": 2.3925,
+      "step": 8645
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998701079210505,
+      "loss": 2.7546,
+      "step": 8650
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998699577161318,
+      "loss": 2.5315,
+      "step": 8655
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998698074244218,
+      "loss": 2.5595,
+      "step": 8660
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998696570459204,
+      "loss": 2.4245,
+      "step": 8665
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998695065806277,
+      "loss": 2.5762,
+      "step": 8670
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999869356028544,
+      "loss": 2.2482,
+      "step": 8675
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999869205389669,
+      "loss": 2.6133,
+      "step": 8680
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998690546640022,
+      "loss": 2.3625,
+      "step": 8685
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998689038515445,
+      "loss": 2.3822,
+      "step": 8690
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998687529522956,
+      "loss": 2.3996,
+      "step": 8695
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998686019662552,
+      "loss": 2.3756,
+      "step": 8700
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998684508934237,
+      "loss": 2.4049,
+      "step": 8705
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998682997338012,
+      "loss": 2.6368,
+      "step": 8710
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999868148487387,
+      "loss": 2.585,
+      "step": 8715
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998679971541821,
+      "loss": 2.4734,
+      "step": 8720
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999867845734186,
+      "loss": 2.5682,
+      "step": 8725
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998676942273986,
+      "loss": 2.2908,
+      "step": 8730
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998675426338204,
+      "loss": 2.3191,
+      "step": 8735
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998673909534507,
+      "loss": 2.4821,
+      "step": 8740
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998672391862896,
+      "loss": 2.5785,
+      "step": 8745
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999867087332338,
+      "loss": 2.5052,
+      "step": 8750
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998669353915948,
+      "loss": 2.5428,
+      "step": 8755
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999866783364061,
+      "loss": 2.6008,
+      "step": 8760
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998666312497359,
+      "loss": 2.2407,
+      "step": 8765
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998664790486196,
+      "loss": 2.3564,
+      "step": 8770
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998663267607126,
+      "loss": 2.4997,
+      "step": 8775
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998661743860142,
+      "loss": 2.613,
+      "step": 8780
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999866021924525,
+      "loss": 2.588,
+      "step": 8785
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998658693762449,
+      "loss": 2.1804,
+      "step": 8790
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999865716741174,
+      "loss": 2.3337,
+      "step": 8795
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998655640193115,
+      "loss": 2.9856,
+      "step": 8800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998654112106582,
+      "loss": 2.5314,
+      "step": 8805
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998652583152145,
+      "loss": 2.454,
+      "step": 8810
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998651053329793,
+      "loss": 2.3367,
+      "step": 8815
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998649522639535,
+      "loss": 2.497,
+      "step": 8820
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998647991081368,
+      "loss": 2.328,
+      "step": 8825
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998646458655288,
+      "loss": 2.4763,
+      "step": 8830
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998644925361302,
+      "loss": 2.6595,
+      "step": 8835
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998643391199407,
+      "loss": 2.4011,
+      "step": 8840
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998641856169605,
+      "loss": 2.2727,
+      "step": 8845
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998640320271894,
+      "loss": 2.5044,
+      "step": 8850
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998638783506274,
+      "loss": 2.5862,
+      "step": 8855
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998637245872747,
+      "loss": 2.5068,
+      "step": 8860
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999863570737131,
+      "loss": 2.6231,
+      "step": 8865
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998634168001965,
+      "loss": 2.2372,
+      "step": 8870
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998632627764716,
+      "loss": 2.4787,
+      "step": 8875
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998631086659557,
+      "loss": 2.3037,
+      "step": 8880
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999862954468649,
+      "loss": 2.4987,
+      "step": 8885
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998628001845515,
+      "loss": 2.4102,
+      "step": 8890
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998626458136634,
+      "loss": 2.4029,
+      "step": 8895
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998624913559847,
+      "loss": 2.4455,
+      "step": 8900
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999862336811515,
+      "loss": 2.7084,
+      "step": 8905
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999862182180255,
+      "loss": 2.3189,
+      "step": 8910
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999862027462204,
+      "loss": 2.3658,
+      "step": 8915
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998618726573626,
+      "loss": 2.3893,
+      "step": 8920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998617177657306,
+      "loss": 2.6753,
+      "step": 8925
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998615627873077,
+      "loss": 2.3598,
+      "step": 8930
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998614077220946,
+      "loss": 2.4231,
+      "step": 8935
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998612525700904,
+      "loss": 2.5501,
+      "step": 8940
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999861097331296,
+      "loss": 2.1479,
+      "step": 8945
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999860942005711,
+      "loss": 2.3937,
+      "step": 8950
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999860786593335,
+      "loss": 2.4062,
+      "step": 8955
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999860631094169,
+      "loss": 2.3751,
+      "step": 8960
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998604755082122,
+      "loss": 2.4611,
+      "step": 8965
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999860319835465,
+      "loss": 2.5114,
+      "step": 8970
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999860164075927,
+      "loss": 2.4888,
+      "step": 8975
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999860008229599,
+      "loss": 2.4488,
+      "step": 8980
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998598522964802,
+      "loss": 2.4818,
+      "step": 8985
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999859696276571,
+      "loss": 2.2975,
+      "step": 8990
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998595401698714,
+      "loss": 2.3841,
+      "step": 8995
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998593839763812,
+      "loss": 2.3571,
+      "step": 9000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999859227696101,
+      "loss": 2.4845,
+      "step": 9005
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998590713290296,
+      "loss": 2.5427,
+      "step": 9010
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998589148751685,
+      "loss": 2.6918,
+      "step": 9015
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998587583345168,
+      "loss": 2.2234,
+      "step": 9020
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999858601707075,
+      "loss": 2.3232,
+      "step": 9025
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998584449928424,
+      "loss": 2.4575,
+      "step": 9030
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998582881918195,
+      "loss": 2.3962,
+      "step": 9035
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998581313040065,
+      "loss": 2.7066,
+      "step": 9040
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998579743294034,
+      "loss": 2.4354,
+      "step": 9045
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998578172680096,
+      "loss": 2.3431,
+      "step": 9050
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998576601198257,
+      "loss": 2.5614,
+      "step": 9055
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999857502884851,
+      "loss": 2.5538,
+      "step": 9060
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998573455630867,
+      "loss": 2.3665,
+      "step": 9065
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999857188154532,
+      "loss": 2.4864,
+      "step": 9070
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998570306591873,
+      "loss": 2.6007,
+      "step": 9075
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999856873077052,
+      "loss": 2.5006,
+      "step": 9080
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998567154081264,
+      "loss": 2.5271,
+      "step": 9085
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999856557652411,
+      "loss": 2.47,
+      "step": 9090
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999856399809905,
+      "loss": 2.3941,
+      "step": 9095
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998562418806093,
+      "loss": 2.493,
+      "step": 9100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998560838645232,
+      "loss": 2.5728,
+      "step": 9105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998559257616472,
+      "loss": 2.594,
+      "step": 9110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998557675719806,
+      "loss": 2.6892,
+      "step": 9115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998556092955242,
+      "loss": 2.3456,
+      "step": 9120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998554509322776,
+      "loss": 2.5928,
+      "step": 9125
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998552924822412,
+      "loss": 2.6073,
+      "step": 9130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998551339454147,
+      "loss": 2.4351,
+      "step": 9135
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998549753217978,
+      "loss": 2.4864,
+      "step": 9140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999854816611391,
+      "loss": 2.4912,
+      "step": 9145
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998546578141942,
+      "loss": 2.4551,
+      "step": 9150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998544989302072,
+      "loss": 2.4463,
+      "step": 9155
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998543399594304,
+      "loss": 2.3603,
+      "step": 9160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998541809018638,
+      "loss": 2.6566,
+      "step": 9165
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999854021757507,
+      "loss": 2.3337,
+      "step": 9170
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000199985386252636,
+      "loss": 2.2276,
+      "step": 9175
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998537032084234,
+      "loss": 2.2761,
+      "step": 9180
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998535438036968,
+      "loss": 2.4544,
+      "step": 9185
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998533843121804,
+      "loss": 2.6465,
+      "step": 9190
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998532247338742,
+      "loss": 2.7588,
+      "step": 9195
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998530650687778,
+      "loss": 2.5723,
+      "step": 9200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998529053168916,
+      "loss": 2.7165,
+      "step": 9205
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998527454782156,
+      "loss": 2.5715,
+      "step": 9210
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998525855527497,
+      "loss": 2.5548,
+      "step": 9215
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999852425540494,
+      "loss": 2.3141,
+      "step": 9220
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998522654414487,
+      "loss": 2.3339,
+      "step": 9225
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998521052556133,
+      "loss": 2.3205,
+      "step": 9230
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999851944982988,
+      "loss": 2.5238,
+      "step": 9235
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998517846235735,
+      "loss": 2.4086,
+      "step": 9240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998516241773686,
+      "loss": 2.2727,
+      "step": 9245
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999851463644374,
+      "loss": 2.0573,
+      "step": 9250
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000199985130302459,
+      "loss": 2.4938,
+      "step": 9255
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999851142318016,
+      "loss": 2.4737,
+      "step": 9260
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998509815246526,
+      "loss": 2.6544,
+      "step": 9265
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998508206444995,
+      "loss": 2.629,
+      "step": 9270
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998506596775566,
+      "loss": 2.1496,
+      "step": 9275
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999850498623824,
+      "loss": 2.6678,
+      "step": 9280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998503374833018,
+      "loss": 2.4694,
+      "step": 9285
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998501762559897,
+      "loss": 2.5012,
+      "step": 9290
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998500149418882,
+      "loss": 2.6216,
+      "step": 9295
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998498535409972,
+      "loss": 2.6186,
+      "step": 9300
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998496920533163,
+      "loss": 2.3252,
+      "step": 9305
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998495304788462,
+      "loss": 2.4164,
+      "step": 9310
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998493688175864,
+      "loss": 2.4692,
+      "step": 9315
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998492070695369,
+      "loss": 2.5259,
+      "step": 9320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999849045234698,
+      "loss": 2.554,
+      "step": 9325
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998488833130695,
+      "loss": 2.3446,
+      "step": 9330
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998487213046515,
+      "loss": 2.3739,
+      "step": 9335
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999848559209444,
+      "loss": 2.3847,
+      "step": 9340
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999848397027447,
+      "loss": 2.4115,
+      "step": 9345
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998482347586606,
+      "loss": 2.4375,
+      "step": 9350
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998480724030848,
+      "loss": 2.5581,
+      "step": 9355
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998479099607195,
+      "loss": 2.485,
+      "step": 9360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998477474315646,
+      "loss": 2.2013,
+      "step": 9365
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998475848156204,
+      "loss": 2.4933,
+      "step": 9370
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999847422112887,
+      "loss": 2.5338,
+      "step": 9375
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999847259323364,
+      "loss": 2.4976,
+      "step": 9380
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998470964470517,
+      "loss": 2.2246,
+      "step": 9385
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000199984693348395,
+      "loss": 2.7514,
+      "step": 9390
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999846770434059,
+      "loss": 2.3547,
+      "step": 9395
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998466072973788,
+      "loss": 2.674,
+      "step": 9400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998464440739092,
+      "loss": 2.2843,
+      "step": 9405
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998462807636503,
+      "loss": 2.5339,
+      "step": 9410
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998461173666022,
+      "loss": 2.3399,
+      "step": 9415
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998459538827644,
+      "loss": 2.6921,
+      "step": 9420
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998457903121377,
+      "loss": 2.4712,
+      "step": 9425
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998456266547216,
+      "loss": 2.5702,
+      "step": 9430
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998454629105165,
+      "loss": 2.5769,
+      "step": 9435
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999845299079522,
+      "loss": 2.3923,
+      "step": 9440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998451351617384,
+      "loss": 2.5397,
+      "step": 9445
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998449711571657,
+      "loss": 2.5886,
+      "step": 9450
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998448070658036,
+      "loss": 2.388,
+      "step": 9455
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998446428876526,
+      "loss": 2.5055,
+      "step": 9460
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998444786227122,
+      "loss": 2.6634,
+      "step": 9465
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998443142709829,
+      "loss": 2.3508,
+      "step": 9470
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998441498324642,
+      "loss": 2.5842,
+      "step": 9475
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998439853071567,
+      "loss": 2.4448,
+      "step": 9480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998438206950597,
+      "loss": 2.5319,
+      "step": 9485
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998436559961742,
+      "loss": 2.4871,
+      "step": 9490
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998434912104994,
+      "loss": 2.3699,
+      "step": 9495
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998433263380353,
+      "loss": 2.4543,
+      "step": 9500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998431613787825,
+      "loss": 2.3799,
+      "step": 9505
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998429963327406,
+      "loss": 2.5814,
+      "step": 9510
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998428311999097,
+      "loss": 2.5413,
+      "step": 9515
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998426659802897,
+      "loss": 2.4872,
+      "step": 9520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998425006738808,
+      "loss": 2.4769,
+      "step": 9525
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999842335280683,
+      "loss": 2.4815,
+      "step": 9530
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998421698006963,
+      "loss": 2.6861,
+      "step": 9535
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998420042339205,
+      "loss": 2.6536,
+      "step": 9540
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999841838580356,
+      "loss": 2.4387,
+      "step": 9545
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998416728400024,
+      "loss": 2.4469,
+      "step": 9550
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000199984150701286,
+      "loss": 2.4486,
+      "step": 9555
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998413410989287,
+      "loss": 2.4776,
+      "step": 9560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998411750982086,
+      "loss": 2.3943,
+      "step": 9565
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998410090106997,
+      "loss": 2.4478,
+      "step": 9570
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998408428364018,
+      "loss": 2.0753,
+      "step": 9575
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998406765753152,
+      "loss": 2.4195,
+      "step": 9580
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998405102274397,
+      "loss": 2.2805,
+      "step": 9585
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998403437927758,
+      "loss": 2.365,
+      "step": 9590
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999840177271323,
+      "loss": 2.5554,
+      "step": 9595
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999840010663081,
+      "loss": 2.5545,
+      "step": 9600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998398439680505,
+      "loss": 2.5549,
+      "step": 9605
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998396771862313,
+      "loss": 2.3808,
+      "step": 9610
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998395103176236,
+      "loss": 2.6222,
+      "step": 9615
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999839343362227,
+      "loss": 2.2378,
+      "step": 9620
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998391763200417,
+      "loss": 2.4597,
+      "step": 9625
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998390091910677,
+      "loss": 2.3753,
+      "step": 9630
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999838841975305,
+      "loss": 2.4199,
+      "step": 9635
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998386746727538,
+      "loss": 2.436,
+      "step": 9640
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999838507283414,
+      "loss": 2.5025,
+      "step": 9645
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998383398072856,
+      "loss": 2.5538,
+      "step": 9650
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998381722443687,
+      "loss": 2.537,
+      "step": 9655
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999838004594663,
+      "loss": 2.4635,
+      "step": 9660
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998378368581688,
+      "loss": 2.3456,
+      "step": 9665
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998376690348862,
+      "loss": 2.1818,
+      "step": 9670
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998375011248147,
+      "loss": 2.6232,
+      "step": 9675
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999837333127955,
+      "loss": 2.4076,
+      "step": 9680
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998371650443067,
+      "loss": 2.4803,
+      "step": 9685
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998369968738698,
+      "loss": 2.3332,
+      "step": 9690
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998368286166444,
+      "loss": 2.4951,
+      "step": 9695
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998366602726306,
+      "loss": 2.562,
+      "step": 9700
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998364918418285,
+      "loss": 2.4566,
+      "step": 9705
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999836323324238,
+      "loss": 2.3356,
+      "step": 9710
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998361547198586,
+      "loss": 2.4319,
+      "step": 9715
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998359860286914,
+      "loss": 2.5457,
+      "step": 9720
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998358172507353,
+      "loss": 2.6287,
+      "step": 9725
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998356483859914,
+      "loss": 2.3725,
+      "step": 9730
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998354794344584,
+      "loss": 2.4876,
+      "step": 9735
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998353103961377,
+      "loss": 2.5241,
+      "step": 9740
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998351412710283,
+      "loss": 2.5487,
+      "step": 9745
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999834972059131,
+      "loss": 2.4608,
+      "step": 9750
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999834802760445,
+      "loss": 2.5501,
+      "step": 9755
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998346333749707,
+      "loss": 2.4901,
+      "step": 9760
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998344639027082,
+      "loss": 2.5478,
+      "step": 9765
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998342943436577,
+      "loss": 2.5238,
+      "step": 9770
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998341246978187,
+      "loss": 2.4842,
+      "step": 9775
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998339549651915,
+      "loss": 2.5017,
+      "step": 9780
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998337851457763,
+      "loss": 2.3788,
+      "step": 9785
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998336152395724,
+      "loss": 2.3579,
+      "step": 9790
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998334452465808,
+      "loss": 2.4703,
+      "step": 9795
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998332751668008,
+      "loss": 2.4456,
+      "step": 9800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998331050002328,
+      "loss": 2.4326,
+      "step": 9805
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998329347468765,
+      "loss": 2.5756,
+      "step": 9810
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999832764406732,
+      "loss": 2.668,
+      "step": 9815
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998325939797994,
+      "loss": 2.3295,
+      "step": 9820
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998324234660788,
+      "loss": 2.4382,
+      "step": 9825
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998322528655703,
+      "loss": 2.2991,
+      "step": 9830
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998320821782738,
+      "loss": 2.7737,
+      "step": 9835
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998319114041888,
+      "loss": 2.3518,
+      "step": 9840
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999831740543316,
+      "loss": 2.2385,
+      "step": 9845
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999831569595655,
+      "loss": 2.503,
+      "step": 9850
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998313985612062,
+      "loss": 2.2877,
+      "step": 9855
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998312274399695,
+      "loss": 2.4239,
+      "step": 9860
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998310562319445,
+      "loss": 2.4293,
+      "step": 9865
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998308849371318,
+      "loss": 2.3999,
+      "step": 9870
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999830713555531,
+      "loss": 2.3348,
+      "step": 9875
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998305420871424,
+      "loss": 2.3459,
+      "step": 9880
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999830370531966,
+      "loss": 2.4825,
+      "step": 9885
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998301988900013,
+      "loss": 2.4744,
+      "step": 9890
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998300271612492,
+      "loss": 2.4583,
+      "step": 9895
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998298553457086,
+      "loss": 2.3683,
+      "step": 9900
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998296834433806,
+      "loss": 2.6798,
+      "step": 9905
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998295114542647,
+      "loss": 2.5938,
+      "step": 9910
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998293393783608,
+      "loss": 2.1591,
+      "step": 9915
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998291672156693,
+      "loss": 2.5002,
+      "step": 9920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000199982899496619,
+      "loss": 2.4791,
+      "step": 9925
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999828822629923,
+      "loss": 2.3907,
+      "step": 9930
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999828650206868,
+      "loss": 2.3974,
+      "step": 9935
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999828477697025,
+      "loss": 2.4692,
+      "step": 9940
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998283051003948,
+      "loss": 2.7643,
+      "step": 9945
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998281324169764,
+      "loss": 2.0845,
+      "step": 9950
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998279596467706,
+      "loss": 2.4465,
+      "step": 9955
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999827786789777,
+      "loss": 2.3468,
+      "step": 9960
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998276138459957,
+      "loss": 2.5807,
+      "step": 9965
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999827440815427,
+      "loss": 2.3106,
+      "step": 9970
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998272676980704,
+      "loss": 2.3665,
+      "step": 9975
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998270944939262,
+      "loss": 2.7522,
+      "step": 9980
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998269212029946,
+      "loss": 2.5925,
+      "step": 9985
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998267478252749,
+      "loss": 2.7227,
+      "step": 9990
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999826574360768,
+      "loss": 2.5847,
+      "step": 9995
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998264008094734,
+      "loss": 2.6716,
+      "step": 10000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998262271713914,
+      "loss": 2.3696,
+      "step": 10005
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998260534465217,
+      "loss": 2.2694,
+      "step": 10010
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998258796348643,
+      "loss": 2.501,
+      "step": 10015
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998257057364196,
+      "loss": 2.4445,
+      "step": 10020
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998255317511875,
+      "loss": 2.2597,
+      "step": 10025
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998253576791676,
+      "loss": 2.5403,
+      "step": 10030
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998251835203604,
+      "loss": 2.4338,
+      "step": 10035
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998250092747658,
+      "loss": 2.4408,
+      "step": 10040
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999824834942384,
+      "loss": 2.3402,
+      "step": 10045
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998246605232143,
+      "loss": 2.2358,
+      "step": 10050
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998244860172575,
+      "loss": 2.4517,
+      "step": 10055
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998243114245133,
+      "loss": 2.2565,
+      "step": 10060
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998241367449814,
+      "loss": 2.5145,
+      "step": 10065
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998239619786626,
+      "loss": 2.2973,
+      "step": 10070
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998237871255562,
+      "loss": 2.4887,
+      "step": 10075
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998236121856623,
+      "loss": 2.2674,
+      "step": 10080
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998234371589813,
+      "loss": 2.472,
+      "step": 10085
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999823262045513,
+      "loss": 2.3701,
+      "step": 10090
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998230868452575,
+      "loss": 2.3732,
+      "step": 10095
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998229115582148,
+      "loss": 2.5064,
+      "step": 10100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998227361843848,
+      "loss": 2.4974,
+      "step": 10105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999822560723767,
+      "loss": 2.3062,
+      "step": 10110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998223851763625,
+      "loss": 2.6467,
+      "step": 10115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998222095421708,
+      "loss": 2.5184,
+      "step": 10120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998220338211917,
+      "loss": 2.5022,
+      "step": 10125
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998218580134257,
+      "loss": 2.5341,
+      "step": 10130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998216821188723,
+      "loss": 2.5238,
+      "step": 10135
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998215061375318,
+      "loss": 2.554,
+      "step": 10140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998213300694042,
+      "loss": 2.1777,
+      "step": 10145
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998211539144894,
+      "loss": 2.4832,
+      "step": 10150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998209776727875,
+      "loss": 2.2942,
+      "step": 10155
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998208013442985,
+      "loss": 2.1188,
+      "step": 10160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998206249290226,
+      "loss": 2.1989,
+      "step": 10165
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998204484269593,
+      "loss": 2.6481,
+      "step": 10170
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998202718381094,
+      "loss": 2.5382,
+      "step": 10175
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998200951624722,
+      "loss": 2.5676,
+      "step": 10180
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999819918400048,
+      "loss": 2.5093,
+      "step": 10185
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998197415508368,
+      "loss": 2.4684,
+      "step": 10190
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999819564614839,
+      "loss": 2.6515,
+      "step": 10195
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998193875920536,
+      "loss": 2.6088,
+      "step": 10200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998192104824818,
+      "loss": 2.2971,
+      "step": 10205
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998190332861225,
+      "loss": 2.5469,
+      "step": 10210
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999818856002977,
+      "loss": 2.294,
+      "step": 10215
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999818678633044,
+      "loss": 2.6456,
+      "step": 10220
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998185011763244,
+      "loss": 2.3013,
+      "step": 10225
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999818323632818,
+      "loss": 2.3921,
+      "step": 10230
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998181460025241,
+      "loss": 2.5392,
+      "step": 10235
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999817968285444,
+      "loss": 2.3093,
+      "step": 10240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999817790481577,
+      "loss": 2.5359,
+      "step": 10245
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999817612590923,
+      "loss": 2.4225,
+      "step": 10250
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998174346134824,
+      "loss": 2.4767,
+      "step": 10255
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999817256549255,
+      "loss": 2.5893,
+      "step": 10260
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998170783982406,
+      "loss": 2.2597,
+      "step": 10265
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998169001604396,
+      "loss": 2.3343,
+      "step": 10270
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999816721835852,
+      "loss": 2.351,
+      "step": 10275
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998165434244775,
+      "loss": 2.4915,
+      "step": 10280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998163649263164,
+      "loss": 2.7405,
+      "step": 10285
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998161863413685,
+      "loss": 2.4724,
+      "step": 10290
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999816007669634,
+      "loss": 2.3576,
+      "step": 10295
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998158289111128,
+      "loss": 2.3789,
+      "step": 10300
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999815650065805,
+      "loss": 2.2525,
+      "step": 10305
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998154711337108,
+      "loss": 2.3558,
+      "step": 10310
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998152921148297,
+      "loss": 2.5284,
+      "step": 10315
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999815113009162,
+      "loss": 2.4762,
+      "step": 10320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999814933816708,
+      "loss": 2.3437,
+      "step": 10325
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999814754537467,
+      "loss": 2.5542,
+      "step": 10330
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998145751714397,
+      "loss": 2.571,
+      "step": 10335
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998143957186256,
+      "loss": 2.3391,
+      "step": 10340
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998142161790255,
+      "loss": 2.418,
+      "step": 10345
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998140365526385,
+      "loss": 2.5464,
+      "step": 10350
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998138568394654,
+      "loss": 2.2553,
+      "step": 10355
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998136770395052,
+      "loss": 2.5427,
+      "step": 10360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998134971527593,
+      "loss": 2.4642,
+      "step": 10365
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998133171792264,
+      "loss": 2.2765,
+      "step": 10370
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999813137118907,
+      "loss": 2.2825,
+      "step": 10375
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998129569718016,
+      "loss": 2.8506,
+      "step": 10380
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998127767379095,
+      "loss": 2.453,
+      "step": 10385
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998125964172314,
+      "loss": 2.2256,
+      "step": 10390
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998124160097667,
+      "loss": 2.5469,
+      "step": 10395
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998122355155155,
+      "loss": 2.7639,
+      "step": 10400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998120549344784,
+      "loss": 2.3243,
+      "step": 10405
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998118742666545,
+      "loss": 2.281,
+      "step": 10410
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998116935120446,
+      "loss": 2.5782,
+      "step": 10415
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998115126706485,
+      "loss": 2.4022,
+      "step": 10420
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999811331742466,
+      "loss": 2.5827,
+      "step": 10425
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999811150727497,
+      "loss": 2.3773,
+      "step": 10430
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999810969625742,
+      "loss": 2.5919,
+      "step": 10435
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999810788437201,
+      "loss": 2.5754,
+      "step": 10440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998106071618735,
+      "loss": 2.5282,
+      "step": 10445
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998104257997598,
+      "loss": 2.4171,
+      "step": 10450
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998102443508602,
+      "loss": 2.3195,
+      "step": 10455
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998100628151742,
+      "loss": 2.4589,
+      "step": 10460
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998098811927018,
+      "loss": 2.4459,
+      "step": 10465
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999809699483444,
+      "loss": 2.4079,
+      "step": 10470
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998095176873996,
+      "loss": 2.5142,
+      "step": 10475
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999809335804569,
+      "loss": 2.4839,
+      "step": 10480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999809153834953,
+      "loss": 2.3362,
+      "step": 10485
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998089717785503,
+      "loss": 2.2893,
+      "step": 10490
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998087896353614,
+      "loss": 2.2023,
+      "step": 10495
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998086074053868,
+      "loss": 2.4619,
+      "step": 10500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998084250886264,
+      "loss": 2.6596,
+      "step": 10505
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998082426850796,
+      "loss": 2.2436,
+      "step": 10510
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999808060194747,
+      "loss": 2.3952,
+      "step": 10515
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998078776176285,
+      "loss": 2.2106,
+      "step": 10520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999807694953724,
+      "loss": 2.4022,
+      "step": 10525
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998075122030335,
+      "loss": 2.5848,
+      "step": 10530
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999807329365557,
+      "loss": 2.5176,
+      "step": 10535
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998071464412948,
+      "loss": 2.4297,
+      "step": 10540
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998069634302468,
+      "loss": 2.4786,
+      "step": 10545
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998067803324125,
+      "loss": 2.2828,
+      "step": 10550
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998065971477927,
+      "loss": 2.2195,
+      "step": 10555
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999806413876387,
+      "loss": 2.6033,
+      "step": 10560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998062305181952,
+      "loss": 2.1797,
+      "step": 10565
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998060470732178,
+      "loss": 2.3711,
+      "step": 10570
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998058635414546,
+      "loss": 2.1863,
+      "step": 10575
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999805679922906,
+      "loss": 2.5752,
+      "step": 10580
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998054962175712,
+      "loss": 2.5801,
+      "step": 10585
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998053124254506,
+      "loss": 2.4377,
+      "step": 10590
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998051285465443,
+      "loss": 2.7624,
+      "step": 10595
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998049445808525,
+      "loss": 2.5697,
+      "step": 10600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998047605283752,
+      "loss": 2.5927,
+      "step": 10605
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999804576389112,
+      "loss": 2.5098,
+      "step": 10610
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998043921630628,
+      "loss": 2.4638,
+      "step": 10615
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998042078502284,
+      "loss": 2.3502,
+      "step": 10620
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999804023450608,
+      "loss": 2.2765,
+      "step": 10625
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999803838964202,
+      "loss": 2.5749,
+      "step": 10630
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999803654391011,
+      "loss": 2.3107,
+      "step": 10635
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998034697310337,
+      "loss": 2.5372,
+      "step": 10640
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998032849842713,
+      "loss": 2.2403,
+      "step": 10645
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998031001507228,
+      "loss": 2.4946,
+      "step": 10650
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998029152303894,
+      "loss": 2.3198,
+      "step": 10655
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998027302232702,
+      "loss": 2.5626,
+      "step": 10660
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998025451293657,
+      "loss": 2.1714,
+      "step": 10665
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998023599486752,
+      "loss": 2.3108,
+      "step": 10670
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998021746811997,
+      "loss": 2.4947,
+      "step": 10675
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998019893269385,
+      "loss": 2.319,
+      "step": 10680
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998018038858918,
+      "loss": 2.1148,
+      "step": 10685
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000199980161835806,
+      "loss": 2.325,
+      "step": 10690
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998014327434426,
+      "loss": 2.4553,
+      "step": 10695
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000199980124704204,
+      "loss": 2.5565,
+      "step": 10700
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999801061253852,
+      "loss": 2.5449,
+      "step": 10705
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998008753788785,
+      "loss": 2.6652,
+      "step": 10710
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998006894171195,
+      "loss": 2.3767,
+      "step": 10715
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998005033685753,
+      "loss": 2.4926,
+      "step": 10720
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999800317233246,
+      "loss": 2.1747,
+      "step": 10725
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999800131011131,
+      "loss": 2.5619,
+      "step": 10730
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999799944702231,
+      "loss": 2.6276,
+      "step": 10735
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997997583065457,
+      "loss": 2.4658,
+      "step": 10740
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997995718240753,
+      "loss": 2.5733,
+      "step": 10745
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997993852548192,
+      "loss": 2.5212,
+      "step": 10750
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997991985987783,
+      "loss": 2.4937,
+      "step": 10755
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999799011855952,
+      "loss": 2.415,
+      "step": 10760
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997988250263407,
+      "loss": 2.3701,
+      "step": 10765
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997986381099442,
+      "loss": 2.5911,
+      "step": 10770
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997984511067624,
+      "loss": 2.7455,
+      "step": 10775
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997982640167954,
+      "loss": 2.4073,
+      "step": 10780
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997980768400434,
+      "loss": 2.5364,
+      "step": 10785
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997978895765063,
+      "loss": 2.5758,
+      "step": 10790
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997977022261844,
+      "loss": 2.4763,
+      "step": 10795
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997975147890772,
+      "loss": 2.3824,
+      "step": 10800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999797327265185,
+      "loss": 2.384,
+      "step": 10805
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997971396545076,
+      "loss": 2.4721,
+      "step": 10810
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997969519570453,
+      "loss": 2.8403,
+      "step": 10815
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999796764172798,
+      "loss": 2.3952,
+      "step": 10820
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997965763017658,
+      "loss": 2.5735,
+      "step": 10825
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997963883439487,
+      "loss": 2.4688,
+      "step": 10830
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997962002993463,
+      "loss": 2.6084,
+      "step": 10835
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997960121679592,
+      "loss": 2.4667,
+      "step": 10840
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997958239497868,
+      "loss": 2.7458,
+      "step": 10845
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000199979563564483,
+      "loss": 2.4085,
+      "step": 10850
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999795447253088,
+      "loss": 2.5596,
+      "step": 10855
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997952587745614,
+      "loss": 2.592,
+      "step": 10860
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997950702092498,
+      "loss": 2.46,
+      "step": 10865
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997948815571532,
+      "loss": 2.5783,
+      "step": 10870
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997946928182721,
+      "loss": 2.441,
+      "step": 10875
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997945039926062,
+      "loss": 2.4168,
+      "step": 10880
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999794315080155,
+      "loss": 2.2499,
+      "step": 10885
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997941260809196,
+      "loss": 2.0727,
+      "step": 10890
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999793936994899,
+      "loss": 2.3397,
+      "step": 10895
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999793747822094,
+      "loss": 2.3988,
+      "step": 10900
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997935585625042,
+      "loss": 2.5731,
+      "step": 10905
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997933692161298,
+      "loss": 2.4224,
+      "step": 10910
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997931797829704,
+      "loss": 2.4046,
+      "step": 10915
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997929902630267,
+      "loss": 2.6267,
+      "step": 10920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999792800656298,
+      "loss": 2.4695,
+      "step": 10925
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997926109627847,
+      "loss": 2.3921,
+      "step": 10930
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997924211824869,
+      "loss": 2.2779,
+      "step": 10935
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997922313154043,
+      "loss": 2.6126,
+      "step": 10940
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997920413615373,
+      "loss": 2.5356,
+      "step": 10945
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997918513208856,
+      "loss": 2.5047,
+      "step": 10950
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997916611934496,
+      "loss": 2.5633,
+      "step": 10955
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997914709792288,
+      "loss": 2.667,
+      "step": 10960
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997912806782236,
+      "loss": 2.4727,
+      "step": 10965
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997910902904337,
+      "loss": 2.4113,
+      "step": 10970
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997908998158593,
+      "loss": 2.6202,
+      "step": 10975
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997907092545006,
+      "loss": 2.5698,
+      "step": 10980
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997905186063574,
+      "loss": 2.6121,
+      "step": 10985
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997903278714299,
+      "loss": 2.6337,
+      "step": 10990
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997901370497179,
+      "loss": 2.4949,
+      "step": 10995
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997899461412212,
+      "loss": 2.2693,
+      "step": 11000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997897551459403,
+      "loss": 2.5515,
+      "step": 11005
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999789564063875,
+      "loss": 2.2905,
+      "step": 11010
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997893728950254,
+      "loss": 2.5977,
+      "step": 11015
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997891816393915,
+      "loss": 2.5307,
+      "step": 11020
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999788990296973,
+      "loss": 2.2319,
+      "step": 11025
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997887988677706,
+      "loss": 2.1402,
+      "step": 11030
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997886073517838,
+      "loss": 2.4426,
+      "step": 11035
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997884157490125,
+      "loss": 2.3057,
+      "step": 11040
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999788224059457,
+      "loss": 2.4781,
+      "step": 11045
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997880322831173,
+      "loss": 2.2988,
+      "step": 11050
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997878404199934,
+      "loss": 2.4854,
+      "step": 11055
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997876484700852,
+      "loss": 2.6659,
+      "step": 11060
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999787456433393,
+      "loss": 2.5361,
+      "step": 11065
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997872643099168,
+      "loss": 2.2406,
+      "step": 11070
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999787072099656,
+      "loss": 2.5424,
+      "step": 11075
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997868798026113,
+      "loss": 2.4807,
+      "step": 11080
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997866874187824,
+      "loss": 2.4698,
+      "step": 11085
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999786494948169,
+      "loss": 2.7042,
+      "step": 11090
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999786302390772,
+      "loss": 2.5465,
+      "step": 11095
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999786109746591,
+      "loss": 2.396,
+      "step": 11100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997859170156255,
+      "loss": 2.3966,
+      "step": 11105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999785724197876,
+      "loss": 2.2927,
+      "step": 11110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999785531293343,
+      "loss": 2.336,
+      "step": 11115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997853383020252,
+      "loss": 2.5682,
+      "step": 11120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997851452239243,
+      "loss": 2.5712,
+      "step": 11125
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999784952059039,
+      "loss": 2.6416,
+      "step": 11130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997847588073696,
+      "loss": 2.3506,
+      "step": 11135
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997845654689162,
+      "loss": 2.5925,
+      "step": 11140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999784372043679,
+      "loss": 2.473,
+      "step": 11145
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999784178531658,
+      "loss": 2.498,
+      "step": 11150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997839849328531,
+      "loss": 2.4977,
+      "step": 11155
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997837912472642,
+      "loss": 2.4077,
+      "step": 11160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997835974748917,
+      "loss": 2.3962,
+      "step": 11165
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999783403615735,
+      "loss": 2.4046,
+      "step": 11170
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997832096697946,
+      "loss": 2.5895,
+      "step": 11175
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997830156370702,
+      "loss": 2.4535,
+      "step": 11180
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999782821517562,
+      "loss": 2.7525,
+      "step": 11185
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997826273112707,
+      "loss": 2.3057,
+      "step": 11190
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997824330181952,
+      "loss": 2.6935,
+      "step": 11195
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999782238638336,
+      "loss": 2.5052,
+      "step": 11200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997820441716927,
+      "loss": 2.3999,
+      "step": 11205
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999781849618266,
+      "loss": 2.2644,
+      "step": 11210
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997816549780558,
+      "loss": 2.082,
+      "step": 11215
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999781460251062,
+      "loss": 2.5398,
+      "step": 11220
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999781265437284,
+      "loss": 2.6221,
+      "step": 11225
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997810705367228,
+      "loss": 2.4482,
+      "step": 11230
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999780875549378,
+      "loss": 2.7528,
+      "step": 11235
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999780680475249,
+      "loss": 2.4769,
+      "step": 11240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999780485314337,
+      "loss": 2.4324,
+      "step": 11245
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997802900666414,
+      "loss": 2.2279,
+      "step": 11250
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999780094732162,
+      "loss": 2.5746,
+      "step": 11255
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999779899310899,
+      "loss": 2.5469,
+      "step": 11260
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997797038028527,
+      "loss": 2.6787,
+      "step": 11265
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997795082080227,
+      "loss": 2.7183,
+      "step": 11270
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997793125264095,
+      "loss": 2.157,
+      "step": 11275
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997791167580126,
+      "loss": 2.355,
+      "step": 11280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997789209028324,
+      "loss": 2.3888,
+      "step": 11285
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997787249608688,
+      "loss": 2.3385,
+      "step": 11290
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997785289321214,
+      "loss": 2.3096,
+      "step": 11295
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999778332816591,
+      "loss": 2.4811,
+      "step": 11300
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999778136614277,
+      "loss": 2.482,
+      "step": 11305
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997779403251799,
+      "loss": 2.3757,
+      "step": 11310
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999777743949299,
+      "loss": 2.5208,
+      "step": 11315
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997775474866352,
+      "loss": 2.3709,
+      "step": 11320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997773509371878,
+      "loss": 2.5319,
+      "step": 11325
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997771543009573,
+      "loss": 2.4362,
+      "step": 11330
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997769575779431,
+      "loss": 2.4256,
+      "step": 11335
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997767607681462,
+      "loss": 2.4115,
+      "step": 11340
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997765638715657,
+      "loss": 2.3964,
+      "step": 11345
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999776366888202,
+      "loss": 2.581,
+      "step": 11350
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997761698180552,
+      "loss": 2.3648,
+      "step": 11355
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997759726611252,
+      "loss": 2.4494,
+      "step": 11360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997757754174118,
+      "loss": 2.2324,
+      "step": 11365
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997755780869154,
+      "loss": 2.4793,
+      "step": 11370
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999775380669636,
+      "loss": 2.3191,
+      "step": 11375
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997751831655734,
+      "loss": 2.7056,
+      "step": 11380
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997749855747276,
+      "loss": 2.3016,
+      "step": 11385
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997747878970986,
+      "loss": 2.3911,
+      "step": 11390
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997745901326866,
+      "loss": 1.878,
+      "step": 11395
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997743922814919,
+      "loss": 2.6576,
+      "step": 11400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997741943435135,
+      "loss": 2.3337,
+      "step": 11405
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997739963187525,
+      "loss": 2.6586,
+      "step": 11410
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997737982072083,
+      "loss": 2.3893,
+      "step": 11415
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997736000088812,
+      "loss": 2.5075,
+      "step": 11420
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999773401723771,
+      "loss": 2.4445,
+      "step": 11425
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997732033518782,
+      "loss": 2.5811,
+      "step": 11430
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997730048932022,
+      "loss": 2.2631,
+      "step": 11435
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997728063477432,
+      "loss": 2.652,
+      "step": 11440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997726077155013,
+      "loss": 2.4095,
+      "step": 11445
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999772408996477,
+      "loss": 2.626,
+      "step": 11450
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999772210190669,
+      "loss": 2.5359,
+      "step": 11455
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997720112980788,
+      "loss": 2.3061,
+      "step": 11460
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997718123187055,
+      "loss": 2.6007,
+      "step": 11465
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997716132525494,
+      "loss": 2.4207,
+      "step": 11470
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997714140996108,
+      "loss": 2.4945,
+      "step": 11475
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999771214859889,
+      "loss": 2.6721,
+      "step": 11480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997710155333844,
+      "loss": 2.4411,
+      "step": 11485
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999770816120097,
+      "loss": 2.5658,
+      "step": 11490
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997706166200273,
+      "loss": 2.4054,
+      "step": 11495
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997704170331745,
+      "loss": 2.2281,
+      "step": 11500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997702173595394,
+      "loss": 2.5157,
+      "step": 11505
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997700175991212,
+      "loss": 2.4029,
+      "step": 11510
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997698177519205,
+      "loss": 2.429,
+      "step": 11515
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997696178179376,
+      "loss": 2.3668,
+      "step": 11520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997694177971713,
+      "loss": 2.3173,
+      "step": 11525
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997692176896228,
+      "loss": 2.492,
+      "step": 11530
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997690174952917,
+      "loss": 2.5129,
+      "step": 11535
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997688172141779,
+      "loss": 2.5201,
+      "step": 11540
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997686168462818,
+      "loss": 2.5068,
+      "step": 11545
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997684163916026,
+      "loss": 2.2567,
+      "step": 11550
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997682158501414,
+      "loss": 2.4654,
+      "step": 11555
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997680152218975,
+      "loss": 2.536,
+      "step": 11560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997678145068713,
+      "loss": 2.661,
+      "step": 11565
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997676137050623,
+      "loss": 2.5768,
+      "step": 11570
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999767412816471,
+      "loss": 2.3125,
+      "step": 11575
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999767211841097,
+      "loss": 2.4354,
+      "step": 11580
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999767010778941,
+      "loss": 2.784,
+      "step": 11585
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997668096300025,
+      "loss": 2.3771,
+      "step": 11590
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997666083942814,
+      "loss": 2.5635,
+      "step": 11595
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997664070717783,
+      "loss": 2.58,
+      "step": 11600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997662056624925,
+      "loss": 2.3966,
+      "step": 11605
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997660041664247,
+      "loss": 2.3584,
+      "step": 11610
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999765802583574,
+      "loss": 2.7565,
+      "step": 11615
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997656009139417,
+      "loss": 2.7276,
+      "step": 11620
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997653991575266,
+      "loss": 2.3058,
+      "step": 11625
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997651973143295,
+      "loss": 2.3889,
+      "step": 11630
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997649953843502,
+      "loss": 2.3774,
+      "step": 11635
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997647933675885,
+      "loss": 2.6754,
+      "step": 11640
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997645912640447,
+      "loss": 2.4136,
+      "step": 11645
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997643890737186,
+      "loss": 2.4073,
+      "step": 11650
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997641867966105,
+      "loss": 2.4641,
+      "step": 11655
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000199976398443272,
+      "loss": 2.5109,
+      "step": 11660
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997637819820475,
+      "loss": 2.3103,
+      "step": 11665
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997635794445927,
+      "loss": 2.1382,
+      "step": 11670
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999763376820356,
+      "loss": 2.4605,
+      "step": 11675
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997631741093373,
+      "loss": 2.4008,
+      "step": 11680
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997629713115362,
+      "loss": 2.3952,
+      "step": 11685
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997627684269532,
+      "loss": 2.5662,
+      "step": 11690
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997625654555881,
+      "loss": 2.575,
+      "step": 11695
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997623623974414,
+      "loss": 2.4816,
+      "step": 11700
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997621592525121,
+      "loss": 2.7463,
+      "step": 11705
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997619560208014,
+      "loss": 2.6753,
+      "step": 11710
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997617527023085,
+      "loss": 2.5616,
+      "step": 11715
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997615492970336,
+      "loss": 2.4672,
+      "step": 11720
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997613458049766,
+      "loss": 2.3435,
+      "step": 11725
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999761142226138,
+      "loss": 2.561,
+      "step": 11730
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999760938560517,
+      "loss": 2.3806,
+      "step": 11735
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997607348081145,
+      "loss": 2.3155,
+      "step": 11740
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997605309689305,
+      "loss": 2.6049,
+      "step": 11745
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997603270429642,
+      "loss": 2.5219,
+      "step": 11750
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997601230302162,
+      "loss": 2.5796,
+      "step": 11755
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997599189306863,
+      "loss": 2.4046,
+      "step": 11760
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997597147443746,
+      "loss": 2.499,
+      "step": 11765
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997595104712812,
+      "loss": 2.4154,
+      "step": 11770
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999759306111406,
+      "loss": 2.3173,
+      "step": 11775
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997591016647493,
+      "loss": 2.4123,
+      "step": 11780
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997588971313107,
+      "loss": 2.5128,
+      "step": 11785
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997586925110902,
+      "loss": 2.4484,
+      "step": 11790
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997584878040883,
+      "loss": 2.5272,
+      "step": 11795
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997582830103046,
+      "loss": 2.5432,
+      "step": 11800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997580781297392,
+      "loss": 2.4824,
+      "step": 11805
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997578731623924,
+      "loss": 2.2588,
+      "step": 11810
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997576681082637,
+      "loss": 2.3555,
+      "step": 11815
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997574629673537,
+      "loss": 2.2945,
+      "step": 11820
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999757257739662,
+      "loss": 2.2826,
+      "step": 11825
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997570524251887,
+      "loss": 2.5673,
+      "step": 11830
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997568470239338,
+      "loss": 2.445,
+      "step": 11835
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997566415358976,
+      "loss": 2.4448,
+      "step": 11840
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000199975643596108,
+      "loss": 2.4024,
+      "step": 11845
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997562302994805,
+      "loss": 2.3226,
+      "step": 11850
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997560245510996,
+      "loss": 2.4601,
+      "step": 11855
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997558187159376,
+      "loss": 2.444,
+      "step": 11860
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999755612793994,
+      "loss": 2.5258,
+      "step": 11865
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997554067852688,
+      "loss": 2.7061,
+      "step": 11870
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997552006897622,
+      "loss": 2.5543,
+      "step": 11875
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997549945074744,
+      "loss": 2.4321,
+      "step": 11880
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997547882384052,
+      "loss": 2.2429,
+      "step": 11885
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997545818825548,
+      "loss": 2.4161,
+      "step": 11890
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997543754399227,
+      "loss": 2.4542,
+      "step": 11895
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997541689105095,
+      "loss": 2.4715,
+      "step": 11900
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999753962294315,
+      "loss": 2.4635,
+      "step": 11905
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997537555913394,
+      "loss": 2.6533,
+      "step": 11910
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997535488015824,
+      "loss": 2.3862,
+      "step": 11915
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999753341925044,
+      "loss": 2.3898,
+      "step": 11920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997531349617246,
+      "loss": 2.6894,
+      "step": 11925
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999752927911624,
+      "loss": 2.5738,
+      "step": 11930
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999752720774742,
+      "loss": 2.3446,
+      "step": 11935
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997525135510792,
+      "loss": 2.6286,
+      "step": 11940
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997523062406352,
+      "loss": 2.535,
+      "step": 11945
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000199975209884341,
+      "loss": 2.5846,
+      "step": 11950
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997518913594036,
+      "loss": 2.6708,
+      "step": 11955
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997516837886158,
+      "loss": 2.3358,
+      "step": 11960
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997514761310473,
+      "loss": 2.5653,
+      "step": 11965
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999751268386698,
+      "loss": 2.4606,
+      "step": 11970
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997510605555672,
+      "loss": 2.6728,
+      "step": 11975
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997508526376555,
+      "loss": 2.5948,
+      "step": 11980
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997506446329627,
+      "loss": 2.4636,
+      "step": 11985
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997504365414895,
+      "loss": 2.6507,
+      "step": 11990
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997502283632349,
+      "loss": 2.4276,
+      "step": 11995
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999750020098199,
+      "loss": 2.4268,
+      "step": 12000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997498117463826,
+      "loss": 2.3155,
+      "step": 12005
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997496033077853,
+      "loss": 2.5103,
+      "step": 12010
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999749394782407,
+      "loss": 2.5187,
+      "step": 12015
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997491861702477,
+      "loss": 2.5975,
+      "step": 12020
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999748977471308,
+      "loss": 2.4078,
+      "step": 12025
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999748768685587,
+      "loss": 2.3794,
+      "step": 12030
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997485598130855,
+      "loss": 2.3124,
+      "step": 12035
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997483508538027,
+      "loss": 2.5713,
+      "step": 12040
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997481418077397,
+      "loss": 2.1336,
+      "step": 12045
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997479326748957,
+      "loss": 2.7057,
+      "step": 12050
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997477234552709,
+      "loss": 2.4896,
+      "step": 12055
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997475141488656,
+      "loss": 2.3709,
+      "step": 12060
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997473047556793,
+      "loss": 2.1572,
+      "step": 12065
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997470952757122,
+      "loss": 2.5542,
+      "step": 12070
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997468857089646,
+      "loss": 2.4698,
+      "step": 12075
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997466760554366,
+      "loss": 2.4545,
+      "step": 12080
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997464663151275,
+      "loss": 2.4641,
+      "step": 12085
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999746256488038,
+      "loss": 2.153,
+      "step": 12090
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999746046574168,
+      "loss": 2.6339,
+      "step": 12095
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997458365735173,
+      "loss": 2.3097,
+      "step": 12100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997456264860862,
+      "loss": 2.196,
+      "step": 12105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997454163118742,
+      "loss": 2.1926,
+      "step": 12110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999745206050882,
+      "loss": 2.524,
+      "step": 12115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999744995703109,
+      "loss": 2.6172,
+      "step": 12120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997447852685557,
+      "loss": 2.401,
+      "step": 12125
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999744574747222,
+      "loss": 2.5944,
+      "step": 12130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999744364139108,
+      "loss": 2.1631,
+      "step": 12135
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997441534442132,
+      "loss": 2.3836,
+      "step": 12140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999743942662538,
+      "loss": 2.5634,
+      "step": 12145
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997437317940825,
+      "loss": 2.4223,
+      "step": 12150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997435208388465,
+      "loss": 2.21,
+      "step": 12155
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997433097968307,
+      "loss": 2.3197,
+      "step": 12160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999743098668034,
+      "loss": 2.3689,
+      "step": 12165
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997428874524568,
+      "loss": 2.3294,
+      "step": 12170
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997426761500997,
+      "loss": 2.2847,
+      "step": 12175
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999742464760962,
+      "loss": 2.5656,
+      "step": 12180
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997422532850442,
+      "loss": 2.4576,
+      "step": 12185
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997420417223463,
+      "loss": 2.5989,
+      "step": 12190
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997418300728682,
+      "loss": 2.323,
+      "step": 12195
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997416183366093,
+      "loss": 2.3168,
+      "step": 12200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997414065135708,
+      "loss": 2.224,
+      "step": 12205
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999741194603752,
+      "loss": 2.5972,
+      "step": 12210
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997409826071527,
+      "loss": 2.5446,
+      "step": 12215
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997407705237737,
+      "loss": 2.4893,
+      "step": 12220
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999740558353614,
+      "loss": 2.5806,
+      "step": 12225
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999740346096675,
+      "loss": 2.4046,
+      "step": 12230
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997401337529553,
+      "loss": 2.3519,
+      "step": 12235
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997399213224557,
+      "loss": 2.1366,
+      "step": 12240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999739708805176,
+      "loss": 2.3766,
+      "step": 12245
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997394962011165,
+      "loss": 2.6111,
+      "step": 12250
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997392835102766,
+      "loss": 2.2581,
+      "step": 12255
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997390707326572,
+      "loss": 2.5957,
+      "step": 12260
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997388578682574,
+      "loss": 2.5943,
+      "step": 12265
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997386449170778,
+      "loss": 2.7057,
+      "step": 12270
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997384318791182,
+      "loss": 2.3825,
+      "step": 12275
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997382187543787,
+      "loss": 2.5424,
+      "step": 12280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997380055428594,
+      "loss": 2.5994,
+      "step": 12285
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000199973779224456,
+      "loss": 2.4017,
+      "step": 12290
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999737578859481,
+      "loss": 2.4752,
+      "step": 12295
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997373653876218,
+      "loss": 2.5205,
+      "step": 12300
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999737151828983,
+      "loss": 2.3572,
+      "step": 12305
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997369381835644,
+      "loss": 2.5049,
+      "step": 12310
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999736724451366,
+      "loss": 2.3371,
+      "step": 12315
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997365106323877,
+      "loss": 2.3831,
+      "step": 12320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000199973629672663,
+      "loss": 2.3459,
+      "step": 12325
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999736082734092,
+      "loss": 2.6217,
+      "step": 12330
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997358686547746,
+      "loss": 2.597,
+      "step": 12335
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997356544886776,
+      "loss": 2.4469,
+      "step": 12340
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997354402358008,
+      "loss": 2.4035,
+      "step": 12345
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997352258961445,
+      "loss": 2.4356,
+      "step": 12350
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999735011469708,
+      "loss": 2.6113,
+      "step": 12355
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997347969564927,
+      "loss": 2.5397,
+      "step": 12360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999734582356497,
+      "loss": 2.5205,
+      "step": 12365
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997343676697222,
+      "loss": 2.525,
+      "step": 12370
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997341528961676,
+      "loss": 2.4736,
+      "step": 12375
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997339380358335,
+      "loss": 2.4842,
+      "step": 12380
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000199973372308872,
+      "loss": 2.5738,
+      "step": 12385
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999733508054827,
+      "loss": 2.4422,
+      "step": 12390
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997332929341544,
+      "loss": 2.286,
+      "step": 12395
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997330777267024,
+      "loss": 2.5369,
+      "step": 12400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997328624324706,
+      "loss": 2.325,
+      "step": 12405
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997326470514598,
+      "loss": 2.5605,
+      "step": 12410
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997324315836697,
+      "loss": 2.4599,
+      "step": 12415
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997322160290998,
+      "loss": 2.514,
+      "step": 12420
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997320003877506,
+      "loss": 2.2743,
+      "step": 12425
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999731784659622,
+      "loss": 2.4482,
+      "step": 12430
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997315688447142,
+      "loss": 2.308,
+      "step": 12435
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999731352943027,
+      "loss": 2.2427,
+      "step": 12440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997311369545604,
+      "loss": 2.1984,
+      "step": 12445
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997309208793146,
+      "loss": 2.368,
+      "step": 12450
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997307047172895,
+      "loss": 2.2944,
+      "step": 12455
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999730488468485,
+      "loss": 2.4126,
+      "step": 12460
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997302721329015,
+      "loss": 2.3809,
+      "step": 12465
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999730055710539,
+      "loss": 2.4451,
+      "step": 12470
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997298392013967,
+      "loss": 2.4656,
+      "step": 12475
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997296226054756,
+      "loss": 2.53,
+      "step": 12480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997294059227754,
+      "loss": 2.561,
+      "step": 12485
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999729189153296,
+      "loss": 2.2128,
+      "step": 12490
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997289722970373,
+      "loss": 2.2274,
+      "step": 12495
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997287553539996,
+      "loss": 2.5228,
+      "step": 12500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999728538324183,
+      "loss": 2.3405,
+      "step": 12505
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999728321207587,
+      "loss": 2.6303,
+      "step": 12510
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997281040042122,
+      "loss": 2.482,
+      "step": 12515
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997278867140583,
+      "loss": 2.6349,
+      "step": 12520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997276693371253,
+      "loss": 2.4412,
+      "step": 12525
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997274518734136,
+      "loss": 2.4151,
+      "step": 12530
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997272343229226,
+      "loss": 2.4361,
+      "step": 12535
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999727016685653,
+      "loss": 2.2396,
+      "step": 12540
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997267989616042,
+      "loss": 2.5266,
+      "step": 12545
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997265811507763,
+      "loss": 2.4731,
+      "step": 12550
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997263632531698,
+      "loss": 2.4637,
+      "step": 12555
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997261452687844,
+      "loss": 2.4655,
+      "step": 12560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997259271976203,
+      "loss": 2.2793,
+      "step": 12565
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999725709039677,
+      "loss": 2.4703,
+      "step": 12570
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997254907949552,
+      "loss": 2.3971,
+      "step": 12575
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997252724634546,
+      "loss": 2.4912,
+      "step": 12580
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997250540451747,
+      "loss": 2.6492,
+      "step": 12585
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997248355401167,
+      "loss": 2.6137,
+      "step": 12590
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997246169482796,
+      "loss": 2.2331,
+      "step": 12595
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999724398269664,
+      "loss": 2.4599,
+      "step": 12600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997241795042696,
+      "loss": 2.6955,
+      "step": 12605
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997239606520966,
+      "loss": 2.4918,
+      "step": 12610
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997237417131446,
+      "loss": 2.4573,
+      "step": 12615
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997235226874141,
+      "loss": 2.5273,
+      "step": 12620
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997233035749052,
+      "loss": 2.5456,
+      "step": 12625
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997230843756173,
+      "loss": 2.4297,
+      "step": 12630
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997228650895513,
+      "loss": 2.5807,
+      "step": 12635
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997226457167062,
+      "loss": 2.6845,
+      "step": 12640
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001999722426257083,
+      "loss": 2.4075,
+      "step": 12645
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001999722206710681,
+      "loss": 2.4821,
+      "step": 12650
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019997219870775005,
+      "loss": 2.3636,
+      "step": 12655
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019997217673575418,
+      "loss": 2.3911,
+      "step": 12660
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019997215475508042,
+      "loss": 2.5498,
+      "step": 12665
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019997213276572883,
+      "loss": 2.2082,
+      "step": 12670
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001999721107676994,
+      "loss": 2.5635,
+      "step": 12675
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001999720887609921,
+      "loss": 2.2998,
+      "step": 12680
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000199972066745607,
+      "loss": 2.2889,
+      "step": 12685
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019997204472154407,
+      "loss": 2.5273,
+      "step": 12690
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019997202268880329,
+      "loss": 2.4649,
+      "step": 12695
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019997200064738463,
+      "loss": 2.5811,
+      "step": 12700
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001999719785972882,
+      "loss": 2.4679,
+      "step": 12705
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001999719565385139,
+      "loss": 2.5025,
+      "step": 12710
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019997193447106182,
+      "loss": 2.3796,
+      "step": 12715
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019997191239493186,
+      "loss": 2.1805,
+      "step": 12720
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001999718903101241,
+      "loss": 2.5732,
+      "step": 12725
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019997186821663852,
+      "loss": 2.3069,
+      "step": 12730
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001999718461144751,
+      "loss": 2.4859,
+      "step": 12735
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019997182400363387,
+      "loss": 2.6011,
+      "step": 12740
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019997180188411484,
+      "loss": 2.5648,
+      "step": 12745
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019997177975591797,
+      "loss": 2.4831,
+      "step": 12750
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019997175761904333,
+      "loss": 2.3341,
+      "step": 12755
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019997173547349083,
+      "loss": 2.6718,
+      "step": 12760
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019997171331926055,
+      "loss": 2.4975,
+      "step": 12765
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001999716911563524,
+      "loss": 2.6542,
+      "step": 12770
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019997166898476652,
+      "loss": 2.3267,
+      "step": 12775
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019997164680450282,
+      "loss": 2.2833,
+      "step": 12780
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019997162461556133,
+      "loss": 2.4378,
+      "step": 12785
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000199971602417942,
+      "loss": 2.1509,
+      "step": 12790
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019997158021164488,
+      "loss": 2.743,
+      "step": 12795
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019997155799667,
+      "loss": 2.3859,
+      "step": 12800
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 1685984,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "save_steps": 200,
+  "total_flos": 3.8338610650381025e+18,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}