diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,9619 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.23539970133662894,
+  "eval_steps": 500,
+  "global_step": 8000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 5e-06,
+      "loss": 15.2915,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1e-05,
+      "loss": 14.8373,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5e-05,
+      "loss": 13.6805,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2e-05,
+      "loss": 12.2794,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5e-05,
+      "loss": 11.1116,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3e-05,
+      "loss": 10.3463,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.5e-05,
+      "loss": 9.7303,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4e-05,
+      "loss": 9.3143,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.5e-05,
+      "loss": 9.0357,
+      "step": 45
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5e-05,
+      "loss": 8.8606,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.500000000000001e-05,
+      "loss": 8.8283,
+      "step": 55
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6e-05,
+      "loss": 8.778,
+      "step": 60
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.500000000000001e-05,
+      "loss": 8.744,
+      "step": 65
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7e-05,
+      "loss": 8.6542,
+      "step": 70
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 8.6223,
+      "step": 75
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8e-05,
+      "loss": 8.6246,
+      "step": 80
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.5e-05,
+      "loss": 8.5221,
+      "step": 85
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9e-05,
+      "loss": 8.4481,
+      "step": 90
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.5e-05,
+      "loss": 8.2447,
+      "step": 95
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001,
+      "loss": 8.1997,
+      "step": 100
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999999462732701e-05,
+      "loss": 8.0566,
+      "step": 105
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999997850930916e-05,
+      "loss": 7.8909,
+      "step": 110
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999995164594991e-05,
+      "loss": 7.7982,
+      "step": 115
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999991403725507e-05,
+      "loss": 7.6743,
+      "step": 120
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.99998656832327e-05,
+      "loss": 7.4999,
+      "step": 125
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999980658389317e-05,
+      "loss": 7.4115,
+      "step": 130
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999973673924921e-05,
+      "loss": 7.3269,
+      "step": 135
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999965614931583e-05,
+      "loss": 7.0671,
+      "step": 140
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999956481411035e-05,
+      "loss": 7.1088,
+      "step": 145
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999946273365238e-05,
+      "loss": 7.0313,
+      "step": 150
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999934990796387e-05,
+      "loss": 6.9729,
+      "step": 155
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999922633706906e-05,
+      "loss": 6.9425,
+      "step": 160
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999909202099453e-05,
+      "loss": 6.7971,
+      "step": 165
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999894695976912e-05,
+      "loss": 6.8043,
+      "step": 170
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.9998791153424e-05,
+      "loss": 6.6299,
+      "step": 175
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999862460199267e-05,
+      "loss": 6.5049,
+      "step": 180
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999844730551092e-05,
+      "loss": 6.5734,
+      "step": 185
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999825926401686e-05,
+      "loss": 6.4745,
+      "step": 190
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999806047755088e-05,
+      "loss": 6.4418,
+      "step": 195
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999785094615573e-05,
+      "loss": 6.3825,
+      "step": 200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99976306698764e-05,
+      "loss": 6.4622,
+      "step": 205
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999739964876026e-05,
+      "loss": 6.4472,
+      "step": 210
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999715788285695e-05,
+      "loss": 6.2238,
+      "step": 215
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999690537221844e-05,
+      "loss": 6.2499,
+      "step": 220
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999664211689897e-05,
+      "loss": 6.3734,
+      "step": 225
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999636811695513e-05,
+      "loss": 6.1738,
+      "step": 230
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99960833724458e-05,
+      "loss": 6.1835,
+      "step": 235
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999578788343218e-05,
+      "loss": 6.1188,
+      "step": 240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999548164997778e-05,
+      "loss": 6.1485,
+      "step": 245
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999516467214839e-05,
+      "loss": 5.9757,
+      "step": 250
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999483695001214e-05,
+      "loss": 6.0638,
+      "step": 255
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999449848363946e-05,
+      "loss": 5.961,
+      "step": 260
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999414927310309e-05,
+      "loss": 5.975,
+      "step": 265
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99937893184781e-05,
+      "loss": 5.8929,
+      "step": 270
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99934186198418e-05,
+      "loss": 5.9359,
+      "step": 275
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999303717727389e-05,
+      "loss": 5.9016,
+      "step": 280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999264499085633e-05,
+      "loss": 5.8167,
+      "step": 285
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999224206067341e-05,
+      "loss": 5.8869,
+      "step": 290
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999182838681171e-05,
+      "loss": 5.8762,
+      "step": 295
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999140396936016e-05,
+      "loss": 5.8273,
+      "step": 300
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999096880840994e-05,
+      "loss": 5.8255,
+      "step": 305
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999052290405459e-05,
+      "loss": 5.8236,
+      "step": 310
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999006625638994e-05,
+      "loss": 5.8806,
+      "step": 315
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998959886551411e-05,
+      "loss": 5.8608,
+      "step": 320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998912073152755e-05,
+      "loss": 5.6891,
+      "step": 325
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998863185453302e-05,
+      "loss": 5.7285,
+      "step": 330
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998813223463557e-05,
+      "loss": 5.7973,
+      "step": 335
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998762187194258e-05,
+      "loss": 5.7853,
+      "step": 340
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998710076656373e-05,
+      "loss": 5.7568,
+      "step": 345
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998656891861102e-05,
+      "loss": 5.6944,
+      "step": 350
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998602632819874e-05,
+      "loss": 5.7363,
+      "step": 355
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998547299544348e-05,
+      "loss": 5.5384,
+      "step": 360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99849089204642e-05,
+      "loss": 5.5935,
+      "step": 365
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998433410338206e-05,
+      "loss": 5.6811,
+      "step": 370
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998374854432064e-05,
+      "loss": 5.5229,
+      "step": 375
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998315224340577e-05,
+      "loss": 5.5843,
+      "step": 380
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99825452007656e-05,
+      "loss": 5.4957,
+      "step": 385
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998192741653057e-05,
+      "loss": 5.4334,
+      "step": 390
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998129889083346e-05,
+      "loss": 5.5941,
+      "step": 395
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998065962380936e-05,
+      "loss": 5.4853,
+      "step": 400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998000961559562e-05,
+      "loss": 5.4953,
+      "step": 405
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997934886633195e-05,
+      "loss": 5.4015,
+      "step": 410
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997867737616035e-05,
+      "loss": 5.4356,
+      "step": 415
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997799514522513e-05,
+      "loss": 5.5643,
+      "step": 420
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99773021736729e-05,
+      "loss": 5.3424,
+      "step": 425
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99765984616526e-05,
+      "loss": 5.4506,
+      "step": 430
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997588400931542e-05,
+      "loss": 5.3998,
+      "step": 435
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997515881681496e-05,
+      "loss": 5.5652,
+      "step": 440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.9974422884307e-05,
+      "loss": 5.4247,
+      "step": 445
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997367621194977e-05,
+      "loss": 5.3456,
+      "step": 450
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99729187999037e-05,
+      "loss": 5.2749,
+      "step": 455
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997215064833155e-05,
+      "loss": 5.3128,
+      "step": 460
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99713717573984e-05,
+      "loss": 5.4097,
+      "step": 465
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997058212727169e-05,
+      "loss": 5.32,
+      "step": 470
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996978175812105e-05,
+      "loss": 5.3478,
+      "step": 475
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996897065011852e-05,
+      "loss": 5.3221,
+      "step": 480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996814880343842e-05,
+      "loss": 5.4686,
+      "step": 485
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996731621825736e-05,
+      "loss": 5.3228,
+      "step": 490
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996647289475426e-05,
+      "loss": 5.282,
+      "step": 495
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996561883311036e-05,
+      "loss": 5.2588,
+      "step": 500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996475403350921e-05,
+      "loss": 5.3736,
+      "step": 505
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.996387849613667e-05,
+      "loss": 5.3389,
+      "step": 510
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.996299222118086e-05,
+      "loss": 5.2222,
+      "step": 515
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99620952088323e-05,
+      "loss": 5.2418,
+      "step": 520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.996118745928373e-05,
+      "loss": 5.2544,
+      "step": 525
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.996026897273024e-05,
+      "loss": 5.1454,
+      "step": 530
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.995933974936922e-05,
+      "loss": 5.3297,
+      "step": 535
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.995839978940036e-05,
+      "loss": 5.1674,
+      "step": 540
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.995744909302567e-05,
+      "loss": 5.3242,
+      "step": 545
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.995648766044947e-05,
+      "loss": 5.2134,
+      "step": 550
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.995551549187836e-05,
+      "loss": 5.249,
+      "step": 555
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.995453258752127e-05,
+      "loss": 5.1446,
+      "step": 560
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.995353894758945e-05,
+      "loss": 5.1268,
+      "step": 565
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.995253457229643e-05,
+      "loss": 5.1354,
+      "step": 570
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.995151946185804e-05,
+      "loss": 5.0708,
+      "step": 575
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.995049361649247e-05,
+      "loss": 5.0358,
+      "step": 580
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994945703642013e-05,
+      "loss": 5.2008,
+      "step": 585
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994840972186386e-05,
+      "loss": 5.1578,
+      "step": 590
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994735167304866e-05,
+      "loss": 5.1646,
+      "step": 595
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994628289020195e-05,
+      "loss": 5.104,
+      "step": 600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994520337355344e-05,
+      "loss": 5.1641,
+      "step": 605
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994411312333507e-05,
+      "loss": 5.0159,
+      "step": 610
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99430121397812e-05,
+      "loss": 5.0888,
+      "step": 615
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994190042312839e-05,
+      "loss": 4.9732,
+      "step": 620
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994077797361559e-05,
+      "loss": 5.0572,
+      "step": 625
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.9939644791484e-05,
+      "loss": 5.0422,
+      "step": 630
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993850087697716e-05,
+      "loss": 5.1352,
+      "step": 635
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99373462303409e-05,
+      "loss": 5.1496,
+      "step": 640
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993618085182337e-05,
+      "loss": 5.0736,
+      "step": 645
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993500474167501e-05,
+      "loss": 5.1817,
+      "step": 650
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99338179001486e-05,
+      "loss": 5.0488,
+      "step": 655
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993262032749914e-05,
+      "loss": 5.0021,
+      "step": 660
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993141202398407e-05,
+      "loss": 5.1246,
+      "step": 665
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993019298986301e-05,
+      "loss": 4.905,
+      "step": 670
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992896322539796e-05,
+      "loss": 5.0456,
+      "step": 675
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992772273085319e-05,
+      "loss": 5.1294,
+      "step": 680
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99264715064953e-05,
+      "loss": 4.9721,
+      "step": 685
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992520955259319e-05,
+      "loss": 5.0823,
+      "step": 690
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992393686941808e-05,
+      "loss": 4.9538,
+      "step": 695
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992265345724344e-05,
+      "loss": 4.9737,
+      "step": 700
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99213593163451e-05,
+      "loss": 4.9815,
+      "step": 705
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992005444700121e-05,
+      "loss": 4.9923,
+      "step": 710
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991873884949214e-05,
+      "loss": 4.9709,
+      "step": 715
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991741252410067e-05,
+      "loss": 5.0237,
+      "step": 720
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99160754711118e-05,
+      "loss": 4.9704,
+      "step": 725
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99147276908129e-05,
+      "loss": 4.9168,
+      "step": 730
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991336918349361e-05,
+      "loss": 4.8494,
+      "step": 735
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991199994944586e-05,
+      "loss": 4.8873,
+      "step": 740
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991061998896393e-05,
+      "loss": 4.9236,
+      "step": 745
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99092293023444e-05,
+      "loss": 4.8342,
+      "step": 750
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990782788988611e-05,
+      "loss": 4.9515,
+      "step": 755
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990641575189022e-05,
+      "loss": 4.8468,
+      "step": 760
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990499288866024e-05,
+      "loss": 4.9213,
+      "step": 765
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990355930050194e-05,
+      "loss": 4.9249,
+      "step": 770
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990211498772341e-05,
+      "loss": 4.978,
+      "step": 775
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990065995063506e-05,
+      "loss": 4.8998,
+      "step": 780
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989919418954956e-05,
+      "loss": 4.8787,
+      "step": 785
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989771770478191e-05,
+      "loss": 4.9366,
+      "step": 790
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989623049664944e-05,
+      "loss": 4.8578,
+      "step": 795
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989473256547175e-05,
+      "loss": 4.8077,
+      "step": 800
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989322391157076e-05,
+      "loss": 4.9438,
+      "step": 805
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989170453527068e-05,
+      "loss": 4.9223,
+      "step": 810
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989017443689805e-05,
+      "loss": 4.7779,
+      "step": 815
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988863361678169e-05,
+      "loss": 4.8883,
+      "step": 820
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988708207525272e-05,
+      "loss": 4.8298,
+      "step": 825
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.98855198126446e-05,
+      "loss": 4.8311,
+      "step": 830
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988394682929305e-05,
+      "loss": 4.8455,
+      "step": 835
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988236312553614e-05,
+      "loss": 4.9231,
+      "step": 840
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988076870171419e-05,
+      "loss": 4.8202,
+      "step": 845
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.987916355816988e-05,
+      "loss": 4.89,
+      "step": 850
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.987754769524814e-05,
+      "loss": 4.8421,
+      "step": 855
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.987592111329626e-05,
+      "loss": 4.9337,
+      "step": 860
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.987428381266377e-05,
+      "loss": 4.774,
+      "step": 865
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.987263579370257e-05,
+      "loss": 4.7861,
+      "step": 870
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.987097705676681e-05,
+      "loss": 4.7715,
+      "step": 875
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.986930760221296e-05,
+      "loss": 4.8519,
+      "step": 880
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.986762743039983e-05,
+      "loss": 4.7235,
+      "step": 885
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.986593654168846e-05,
+      "loss": 4.7135,
+      "step": 890
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.986423493644225e-05,
+      "loss": 4.824,
+      "step": 895
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.98625226150269e-05,
+      "loss": 4.8853,
+      "step": 900
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.986079957781038e-05,
+      "loss": 4.8289,
+      "step": 905
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.9859065825163e-05,
+      "loss": 4.7917,
+      "step": 910
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.985732135745733e-05,
+      "loss": 4.8084,
+      "step": 915
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.985556617506828e-05,
+      "loss": 4.6926,
+      "step": 920
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.985380027837308e-05,
+      "loss": 4.7937,
+      "step": 925
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.98520236677512e-05,
+      "loss": 4.7026,
+      "step": 930
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.985023634358443e-05,
+      "loss": 4.7514,
+      "step": 935
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984843830625691e-05,
+      "loss": 4.8102,
+      "step": 940
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984662955615507e-05,
+      "loss": 4.8551,
+      "step": 945
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984481009366758e-05,
+      "loss": 4.7303,
+      "step": 950
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984297991918548e-05,
+      "loss": 4.6998,
+      "step": 955
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984113903310206e-05,
+      "loss": 4.7362,
+      "step": 960
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983928743581296e-05,
+      "loss": 4.7977,
+      "step": 965
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983742512771611e-05,
+      "loss": 4.6277,
+      "step": 970
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983555210921171e-05,
+      "loss": 4.7935,
+      "step": 975
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983366838070231e-05,
+      "loss": 4.7157,
+      "step": 980
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983177394259271e-05,
+      "loss": 4.7415,
+      "step": 985
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.982986879529006e-05,
+      "loss": 4.7386,
+      "step": 990
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.982795293920377e-05,
+      "loss": 4.7188,
+      "step": 995
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.982602637474558e-05,
+      "loss": 4.569,
+      "step": 1000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.982408910232952e-05,
+      "loss": 4.6772,
+      "step": 1005
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.982214112237193e-05,
+      "loss": 4.7104,
+      "step": 1010
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.982018243529146e-05,
+      "loss": 4.645,
+      "step": 1015
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.981821304150901e-05,
+      "loss": 4.7118,
+      "step": 1020
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.981623294144782e-05,
+      "loss": 4.6867,
+      "step": 1025
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.981424213553344e-05,
+      "loss": 4.6214,
+      "step": 1030
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.981224062419372e-05,
+      "loss": 4.6932,
+      "step": 1035
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.981022840785877e-05,
+      "loss": 4.7037,
+      "step": 1040
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.980820548696106e-05,
+      "loss": 4.5934,
+      "step": 1045
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.98061718619353e-05,
+      "loss": 4.6578,
+      "step": 1050
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.980412753321856e-05,
+      "loss": 4.6308,
+      "step": 1055
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.980207250125015e-05,
+      "loss": 4.622,
+      "step": 1060
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.980000676647174e-05,
+      "loss": 4.6667,
+      "step": 1065
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.979793032932725e-05,
+      "loss": 4.7251,
+      "step": 1070
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.979584319026292e-05,
+      "loss": 4.6105,
+      "step": 1075
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.97937453497273e-05,
+      "loss": 4.7102,
+      "step": 1080
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.979163680817124e-05,
+      "loss": 4.6989,
+      "step": 1085
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.978951756604786e-05,
+      "loss": 4.7414,
+      "step": 1090
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.978738762381262e-05,
+      "loss": 4.6414,
+      "step": 1095
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.978524698192322e-05,
+      "loss": 4.6652,
+      "step": 1100
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.978309564083975e-05,
+      "loss": 4.5955,
+      "step": 1105
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.97809336010245e-05,
+      "loss": 4.6192,
+      "step": 1110
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.977876086294216e-05,
+      "loss": 4.4701,
+      "step": 1115
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.977657742705962e-05,
+      "loss": 4.6778,
+      "step": 1120
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.977438329384614e-05,
+      "loss": 4.5762,
+      "step": 1125
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.977217846377324e-05,
+      "loss": 4.5703,
+      "step": 1130
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.976996293731476e-05,
+      "loss": 4.4852,
+      "step": 1135
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.976773671494684e-05,
+      "loss": 4.5364,
+      "step": 1140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.976549979714791e-05,
+      "loss": 4.587,
+      "step": 1145
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.976325218439868e-05,
+      "loss": 4.5353,
+      "step": 1150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.976099387718218e-05,
+      "loss": 4.5999,
+      "step": 1155
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.975872487598377e-05,
+      "loss": 4.4865,
+      "step": 1160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.975644518129104e-05,
+      "loss": 4.5865,
+      "step": 1165
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.975415479359394e-05,
+      "loss": 4.6696,
+      "step": 1170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.975185371338464e-05,
+      "loss": 4.5801,
+      "step": 1175
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.97495419411577e-05,
+      "loss": 4.6272,
+      "step": 1180
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.974721947740995e-05,
+      "loss": 4.5121,
+      "step": 1185
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.974488632264046e-05,
+      "loss": 4.5602,
+      "step": 1190
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.974254247735066e-05,
+      "loss": 4.5405,
+      "step": 1195
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.974018794204426e-05,
+      "loss": 4.5751,
+      "step": 1200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.973782271722727e-05,
+      "loss": 4.6431,
+      "step": 1205
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.973544680340799e-05,
+      "loss": 4.508,
+      "step": 1210
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.973306020109701e-05,
+      "loss": 4.5249,
+      "step": 1215
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.973066291080724e-05,
+      "loss": 4.5344,
+      "step": 1220
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.972825493305387e-05,
+      "loss": 4.5477,
+      "step": 1225
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.97258362683544e-05,
+      "loss": 4.6301,
+      "step": 1230
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.972340691722858e-05,
+      "loss": 4.4456,
+      "step": 1235
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.972096688019856e-05,
+      "loss": 4.4671,
+      "step": 1240
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.971851615778868e-05,
+      "loss": 4.4511,
+      "step": 1245
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.971605475052561e-05,
+      "loss": 4.5892,
+      "step": 1250
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.971358265893834e-05,
+      "loss": 4.6378,
+      "step": 1255
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.971109988355814e-05,
+      "loss": 4.384,
+      "step": 1260
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.970860642491854e-05,
+      "loss": 4.5436,
+      "step": 1265
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.970610228355548e-05,
+      "loss": 4.4915,
+      "step": 1270
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.970358746000704e-05,
+      "loss": 4.4602,
+      "step": 1275
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.97010619548137e-05,
+      "loss": 4.5206,
+      "step": 1280
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.969852576851822e-05,
+      "loss": 4.6151,
+      "step": 1285
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.969597890166565e-05,
+      "loss": 4.5681,
+      "step": 1290
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.969342135480328e-05,
+      "loss": 4.4474,
+      "step": 1295
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.969085312848081e-05,
+      "loss": 4.459,
+      "step": 1300
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.968827422325012e-05,
+      "loss": 4.5095,
+      "step": 1305
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.968568463966545e-05,
+      "loss": 4.3632,
+      "step": 1310
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.968308437828333e-05,
+      "loss": 4.4161,
+      "step": 1315
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.968047343966256e-05,
+      "loss": 4.5433,
+      "step": 1320
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.967785182436426e-05,
+      "loss": 4.562,
+      "step": 1325
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.967521953295185e-05,
+      "loss": 4.5134,
+      "step": 1330
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.967257656599099e-05,
+      "loss": 4.483,
+      "step": 1335
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.966992292404967e-05,
+      "loss": 4.4504,
+      "step": 1340
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.966725860769821e-05,
+      "loss": 4.4237,
+      "step": 1345
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.966458361750916e-05,
+      "loss": 4.4394,
+      "step": 1350
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.966189795405743e-05,
+      "loss": 4.4737,
+      "step": 1355
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.965920161792016e-05,
+      "loss": 4.544,
+      "step": 1360
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.96564946096768e-05,
+      "loss": 4.4541,
+      "step": 1365
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.965377692990914e-05,
+      "loss": 4.4365,
+      "step": 1370
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.965104857920121e-05,
+      "loss": 4.4468,
+      "step": 1375
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.964830955813935e-05,
+      "loss": 4.4687,
+      "step": 1380
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.964555986731219e-05,
+      "loss": 4.3348,
+      "step": 1385
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.964279950731066e-05,
+      "loss": 4.4159,
+      "step": 1390
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.9640028478728e-05,
+      "loss": 4.41,
+      "step": 1395
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.963724678215968e-05,
+      "loss": 4.3369,
+      "step": 1400
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.963445441820356e-05,
+      "loss": 4.4003,
+      "step": 1405
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.963165138745969e-05,
+      "loss": 4.428,
+      "step": 1410
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.962883769053051e-05,
+      "loss": 4.5188,
+      "step": 1415
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.962601332802065e-05,
+      "loss": 4.516,
+      "step": 1420
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.962317830053711e-05,
+      "loss": 4.4646,
+      "step": 1425
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.962033260868916e-05,
+      "loss": 4.2593,
+      "step": 1430
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.961747625308838e-05,
+      "loss": 4.5174,
+      "step": 1435
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.961460923434858e-05,
+      "loss": 4.4841,
+      "step": 1440
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.961173155308591e-05,
+      "loss": 4.4687,
+      "step": 1445
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.960884320991881e-05,
+      "loss": 4.4379,
+      "step": 1450
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.960594420546802e-05,
+      "loss": 4.4335,
+      "step": 1455
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.960303454035654e-05,
+      "loss": 4.4521,
+      "step": 1460
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.960011421520969e-05,
+      "loss": 4.3527,
+      "step": 1465
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.959718323065503e-05,
+      "loss": 4.4126,
+      "step": 1470
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.959424158732252e-05,
+      "loss": 4.3958,
+      "step": 1475
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.959128928584426e-05,
+      "loss": 4.5376,
+      "step": 1480
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.958832632685478e-05,
+      "loss": 4.4431,
+      "step": 1485
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.95853527109908e-05,
+      "loss": 4.4779,
+      "step": 1490
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.95823684388914e-05,
+      "loss": 4.4424,
+      "step": 1495
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.95793735111979e-05,
+      "loss": 4.4295,
+      "step": 1500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.957636792855394e-05,
+      "loss": 4.4732,
+      "step": 1505
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.957335169160545e-05,
+      "loss": 4.306,
+      "step": 1510
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.957032480100061e-05,
+      "loss": 4.4493,
+      "step": 1515
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.956728725738996e-05,
+      "loss": 4.3906,
+      "step": 1520
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.956423906142626e-05,
+      "loss": 4.3141,
+      "step": 1525
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.956118021376461e-05,
+      "loss": 4.374,
+      "step": 1530
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.955811071506236e-05,
+      "loss": 4.2777,
+      "step": 1535
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.955503056597917e-05,
+      "loss": 4.3111,
+      "step": 1540
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.955193976717698e-05,
+      "loss": 4.4216,
+      "step": 1545
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.954883831932003e-05,
+      "loss": 4.3938,
+      "step": 1550
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.954572622307485e-05,
+      "loss": 4.3194,
+      "step": 1555
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.954260347911025e-05,
+      "loss": 4.3174,
+      "step": 1560
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.953947008809731e-05,
+      "loss": 4.3371,
+      "step": 1565
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.953632605070944e-05,
+      "loss": 4.4209,
+      "step": 1570
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.95331713676223e-05,
+      "loss": 4.3032,
+      "step": 1575
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.953000603951386e-05,
+      "loss": 4.4123,
+      "step": 1580
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.952683006706438e-05,
+      "loss": 4.2631,
+      "step": 1585
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.952364345095638e-05,
+      "loss": 4.3549,
+      "step": 1590
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.95204461918747e-05,
+      "loss": 4.368,
+      "step": 1595
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.951723829050646e-05,
+      "loss": 4.3894,
+      "step": 1600
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.951401974754101e-05,
+      "loss": 4.3082,
+      "step": 1605
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.951079056367013e-05,
+      "loss": 4.3826,
+      "step": 1610
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.95075507395877e-05,
+      "loss": 4.3867,
+      "step": 1615
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.950430027599003e-05,
+      "loss": 4.344,
+      "step": 1620
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.950103917357566e-05,
+      "loss": 4.3297,
+      "step": 1625
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.949776743304542e-05,
+      "loss": 4.3006,
+      "step": 1630
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.949448505510243e-05,
+      "loss": 4.2923,
+      "step": 1635
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.94911920404521e-05,
+      "loss": 4.3101,
+      "step": 1640
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.94878883898021e-05,
+      "loss": 4.2658,
+      "step": 1645
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.948457410386244e-05,
+      "loss": 4.3168,
+      "step": 1650
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.948124918334538e-05,
+      "loss": 4.3695,
+      "step": 1655
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.947791362896543e-05,
+      "loss": 4.4382,
+      "step": 1660
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.947456744143947e-05,
+      "loss": 4.481,
+      "step": 1665
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.947121062148658e-05,
+      "loss": 4.3398,
+      "step": 1670
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.94678431698282e-05,
+      "loss": 4.4429,
+      "step": 1675
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.946446508718801e-05,
+      "loss": 4.2906,
+      "step": 1680
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.946107637429196e-05,
+      "loss": 4.3386,
+      "step": 1685
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.945767703186832e-05,
+      "loss": 4.4313,
+      "step": 1690
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.945426706064764e-05,
+      "loss": 4.2688,
+      "step": 1695
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.945084646136275e-05,
+      "loss": 4.4541,
+      "step": 1700
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.944741523474874e-05,
+      "loss": 4.2454,
+      "step": 1705
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.944397338154301e-05,
+      "loss": 4.3106,
+      "step": 1710
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.944052090248526e-05,
+      "loss": 4.2987,
+      "step": 1715
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.943705779831742e-05,
+      "loss": 4.2808,
+      "step": 1720
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.943358406978377e-05,
+      "loss": 4.2904,
+      "step": 1725
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.943009971763082e-05,
+      "loss": 4.3078,
+      "step": 1730
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.942660474260737e-05,
+      "loss": 4.2922,
+      "step": 1735
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.942309914546453e-05,
+      "loss": 4.2959,
+      "step": 1740
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.941958292695565e-05,
+      "loss": 4.3421,
+      "step": 1745
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.941605608783644e-05,
+      "loss": 4.3147,
+      "step": 1750
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.94125186288648e-05,
+      "loss": 4.3549,
+      "step": 1755
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.940897055080098e-05,
+      "loss": 4.3661,
+      "step": 1760
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.940541185440746e-05,
+      "loss": 4.2695,
+      "step": 1765
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.940184254044904e-05,
+      "loss": 4.2835,
+      "step": 1770
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.939826260969282e-05,
+      "loss": 4.3148,
+      "step": 1775
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.93946720629081e-05,
+      "loss": 4.2701,
+      "step": 1780
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.939107090086654e-05,
+      "loss": 4.2671,
+      "step": 1785
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.938745912434208e-05,
+      "loss": 4.2858,
+      "step": 1790
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.938383673411087e-05,
+      "loss": 4.2416,
+      "step": 1795
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.938020373095141e-05,
+      "loss": 4.3953,
+      "step": 1800
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.937656011564445e-05,
+      "loss": 4.2623,
+      "step": 1805
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.937290588897302e-05,
+      "loss": 4.2621,
+      "step": 1810
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.936924105172248e-05,
+      "loss": 4.2236,
+      "step": 1815
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.936556560468037e-05,
+      "loss": 4.2463,
+      "step": 1820
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.93618795486366e-05,
+      "loss": 4.2882,
+      "step": 1825
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.935818288438334e-05,
+      "loss": 4.1092,
+      "step": 1830
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.935447561271501e-05,
+      "loss": 4.3282,
+      "step": 1835
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.935075773442835e-05,
+      "loss": 4.2477,
+      "step": 1840
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.934702925032232e-05,
+      "loss": 4.216,
+      "step": 1845
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.934329016119824e-05,
+      "loss": 4.299,
+      "step": 1850
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.933954046785963e-05,
+      "loss": 4.3033,
+      "step": 1855
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.933578017111233e-05,
+      "loss": 4.4659,
+      "step": 1860
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.933200927176449e-05,
+      "loss": 4.334,
+      "step": 1865
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.932822777062646e-05,
+      "loss": 4.331,
+      "step": 1870
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.932443566851092e-05,
+      "loss": 4.3586,
+      "step": 1875
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.932063296623284e-05,
+      "loss": 4.1845,
+      "step": 1880
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.931681966460944e-05,
+      "loss": 4.2728,
+      "step": 1885
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.93129957644602e-05,
+      "loss": 4.343,
+      "step": 1890
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.930916126660691e-05,
+      "loss": 4.2958,
+      "step": 1895
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.930531617187367e-05,
+      "loss": 4.2894,
+      "step": 1900
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.930146048108678e-05,
+      "loss": 4.2469,
+      "step": 1905
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.929759419507487e-05,
+      "loss": 4.194,
+      "step": 1910
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.92937173146688e-05,
+      "loss": 4.2045,
+      "step": 1915
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.928982984070179e-05,
+      "loss": 4.2511,
+      "step": 1920
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.928593177400925e-05,
+      "loss": 4.2092,
+      "step": 1925
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.928202311542889e-05,
+      "loss": 4.2628,
+      "step": 1930
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.927810386580075e-05,
+      "loss": 4.2659,
+      "step": 1935
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.927417402596707e-05,
+      "loss": 4.2317,
+      "step": 1940
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.927023359677241e-05,
+      "loss": 4.3355,
+      "step": 1945
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.92662825790636e-05,
+      "loss": 4.2747,
+      "step": 1950
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.926232097368974e-05,
+      "loss": 4.3118,
+      "step": 1955
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.925834878150219e-05,
+      "loss": 4.1735,
+      "step": 1960
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.925436600335464e-05,
+      "loss": 4.2208,
+      "step": 1965
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.925037264010298e-05,
+      "loss": 4.1811,
+      "step": 1970
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.924636869260542e-05,
+      "loss": 4.1976,
+      "step": 1975
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.924235416172243e-05,
+      "loss": 4.2036,
+      "step": 1980
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.923832904831679e-05,
+      "loss": 4.0568,
+      "step": 1985
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.923429335325349e-05,
+      "loss": 4.123,
+      "step": 1990
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.923024707739985e-05,
+      "loss": 4.141,
+      "step": 1995
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.922619022162545e-05,
+      "loss": 4.3676,
+      "step": 2000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.92221227868021e-05,
+      "loss": 4.1095,
+      "step": 2005
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.921804477380394e-05,
+      "loss": 4.183,
+      "step": 2010
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.921395618350739e-05,
+      "loss": 4.1026,
+      "step": 2015
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.920985701679106e-05,
+      "loss": 4.1973,
+      "step": 2020
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.920574727453595e-05,
+      "loss": 4.2537,
+      "step": 2025
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.920162695762522e-05,
+      "loss": 4.2212,
+      "step": 2030
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.919749606694439e-05,
+      "loss": 4.2762,
+      "step": 2035
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.919335460338122e-05,
+      "loss": 4.2108,
+      "step": 2040
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.918920256782572e-05,
+      "loss": 4.2716,
+      "step": 2045
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.918503996117016e-05,
+      "loss": 4.1699,
+      "step": 2050
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.918086678430917e-05,
+      "loss": 4.1957,
+      "step": 2055
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.917668303813959e-05,
+      "loss": 4.1147,
+      "step": 2060
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.917248872356051e-05,
+      "loss": 4.2329,
+      "step": 2065
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.916828384147331e-05,
+      "loss": 4.1625,
+      "step": 2070
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.916406839278167e-05,
+      "loss": 4.2083,
+      "step": 2075
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.915984237839152e-05,
+      "loss": 4.2749,
+      "step": 2080
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.915560579921106e-05,
+      "loss": 4.2271,
+      "step": 2085
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.915135865615074e-05,
+      "loss": 4.1895,
+      "step": 2090
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.914710095012332e-05,
+      "loss": 4.2209,
+      "step": 2095
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.91428326820438e-05,
+      "loss": 4.1157,
+      "step": 2100
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.913855385282947e-05,
+      "loss": 4.1636,
+      "step": 2105
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.913426446339986e-05,
+      "loss": 4.0183,
+      "step": 2110
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.912996451467682e-05,
+      "loss": 4.1938,
+      "step": 2115
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.912565400758443e-05,
+      "loss": 4.251,
+      "step": 2120
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.912133294304901e-05,
+      "loss": 4.2339,
+      "step": 2125
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.911700132199924e-05,
+      "loss": 4.0611,
+      "step": 2130
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.911265914536598e-05,
+      "loss": 3.9743,
+      "step": 2135
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.910830641408242e-05,
+      "loss": 4.1893,
+      "step": 2140
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.910394312908397e-05,
+      "loss": 4.1794,
+      "step": 2145
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.909956929130834e-05,
+      "loss": 4.1843,
+      "step": 2150
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.90951849016955e-05,
+      "loss": 4.1173,
+      "step": 2155
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.909078996118768e-05,
+      "loss": 4.0553,
+      "step": 2160
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.908638447072939e-05,
+      "loss": 4.1138,
+      "step": 2165
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.90819684312674e-05,
+      "loss": 4.1983,
+      "step": 2170
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.907754184375074e-05,
+      "loss": 4.1783,
+      "step": 2175
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.90731047091307e-05,
+      "loss": 4.2156,
+      "step": 2180
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.906865702836088e-05,
+      "loss": 4.1763,
+      "step": 2185
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.90641988023971e-05,
+      "loss": 4.1458,
+      "step": 2190
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.905973003219748e-05,
+      "loss": 4.1405,
+      "step": 2195
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.905525071872237e-05,
+      "loss": 4.1137,
+      "step": 2200
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.905076086293443e-05,
+      "loss": 4.0714,
+      "step": 2205
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.904626046579852e-05,
+      "loss": 4.0644,
+      "step": 2210
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.904174952828186e-05,
+      "loss": 4.1475,
+      "step": 2215
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.903722805135385e-05,
+      "loss": 4.1335,
+      "step": 2220
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.903269603598617e-05,
+      "loss": 4.2312,
+      "step": 2225
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.902815348315282e-05,
+      "loss": 4.0862,
+      "step": 2230
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.902360039383e-05,
+      "loss": 4.2002,
+      "step": 2235
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.901903676899621e-05,
+      "loss": 4.1572,
+      "step": 2240
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.901446260963221e-05,
+      "loss": 4.175,
+      "step": 2245
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.9009877916721e-05,
+      "loss": 4.0916,
+      "step": 2250
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.900528269124787e-05,
+      "loss": 4.1309,
+      "step": 2255
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.900067693420039e-05,
+      "loss": 4.1768,
+      "step": 2260
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.899606064656832e-05,
+      "loss": 4.1725,
+      "step": 2265
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.899143382934379e-05,
+      "loss": 4.0711,
+      "step": 2270
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.898679648352108e-05,
+      "loss": 4.0803,
+      "step": 2275
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.898214861009683e-05,
+      "loss": 4.1041,
+      "step": 2280
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.897749021006988e-05,
+      "loss": 4.2048,
+      "step": 2285
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.897282128444135e-05,
+      "loss": 4.1153,
+      "step": 2290
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.896814183421464e-05,
+      "loss": 4.1368,
+      "step": 2295
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.896345186039538e-05,
+      "loss": 4.1931,
+      "step": 2300
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.895875136399149e-05,
+      "loss": 4.1113,
+      "step": 2305
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.895404034601313e-05,
+      "loss": 4.0688,
+      "step": 2310
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.894931880747273e-05,
+      "loss": 4.0385,
+      "step": 2315
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.894458674938499e-05,
+      "loss": 4.0971,
+      "step": 2320
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.893984417276686e-05,
+      "loss": 4.1397,
+      "step": 2325
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.893509107863754e-05,
+      "loss": 4.1938,
+      "step": 2330
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.893032746801854e-05,
+      "loss": 4.2602,
+      "step": 2335
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.892555334193353e-05,
+      "loss": 4.1083,
+      "step": 2340
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.892076870140857e-05,
+      "loss": 3.9812,
+      "step": 2345
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.891597354747186e-05,
+      "loss": 4.0447,
+      "step": 2350
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.891116788115393e-05,
+      "loss": 4.0988,
+      "step": 2355
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.890635170348756e-05,
+      "loss": 4.1969,
+      "step": 2360
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.890152501550777e-05,
+      "loss": 3.9953,
+      "step": 2365
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.889668781825185e-05,
+      "loss": 4.059,
+      "step": 2370
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.889184011275936e-05,
+      "loss": 3.9866,
+      "step": 2375
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.888698190007209e-05,
+      "loss": 4.0989,
+      "step": 2380
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.88821131812341e-05,
+      "loss": 4.1694,
+      "step": 2385
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.887723395729174e-05,
+      "loss": 4.0955,
+      "step": 2390
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.887234422929355e-05,
+      "loss": 4.0064,
+      "step": 2395
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.886744399829039e-05,
+      "loss": 4.0144,
+      "step": 2400
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.886253326533535e-05,
+      "loss": 4.1039,
+      "step": 2405
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.885761203148378e-05,
+      "loss": 4.1546,
+      "step": 2410
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.885268029779327e-05,
+      "loss": 4.054,
+      "step": 2415
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.884773806532373e-05,
+      "loss": 4.1546,
+      "step": 2420
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.884278533513723e-05,
+      "loss": 4.1205,
+      "step": 2425
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.883782210829816e-05,
+      "loss": 4.1159,
+      "step": 2430
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.883284838587317e-05,
+      "loss": 4.1437,
+      "step": 2435
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.882786416893112e-05,
+      "loss": 4.1441,
+      "step": 2440
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.882286945854319e-05,
+      "loss": 4.0811,
+      "step": 2445
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.881786425578274e-05,
+      "loss": 4.0699,
+      "step": 2450
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.881284856172545e-05,
+      "loss": 4.0569,
+      "step": 2455
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.880782237744921e-05,
+      "loss": 4.0931,
+      "step": 2460
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.880278570403419e-05,
+      "loss": 4.1687,
+      "step": 2465
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.87977385425628e-05,
+      "loss": 4.0318,
+      "step": 2470
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.87926808941197e-05,
+      "loss": 4.0412,
+      "step": 2475
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.878761275979184e-05,
+      "loss": 4.1549,
+      "step": 2480
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.878253414066839e-05,
+      "loss": 4.1028,
+      "step": 2485
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.877744503784076e-05,
+      "loss": 4.0762,
+      "step": 2490
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.877234545240265e-05,
+      "loss": 4.0263,
+      "step": 2495
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.876723538545001e-05,
+      "loss": 4.1206,
+      "step": 2500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.8762114838081e-05,
+      "loss": 4.0413,
+      "step": 2505
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.875698381139607e-05,
+      "loss": 3.949,
+      "step": 2510
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.875184230649792e-05,
+      "loss": 4.099,
+      "step": 2515
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.874669032449149e-05,
+      "loss": 4.2385,
+      "step": 2520
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.874152786648399e-05,
+      "loss": 3.9792,
+      "step": 2525
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.873635493358484e-05,
+      "loss": 4.0007,
+      "step": 2530
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.873117152690576e-05,
+      "loss": 4.0371,
+      "step": 2535
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.872597764756071e-05,
+      "loss": 4.0825,
+      "step": 2540
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.872077329666587e-05,
+      "loss": 4.1262,
+      "step": 2545
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.871555847533968e-05,
+      "loss": 4.1624,
+      "step": 2550
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.871033318470289e-05,
+      "loss": 4.0073,
+      "step": 2555
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.870509742587838e-05,
+      "loss": 4.0994,
+      "step": 2560
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.869985119999143e-05,
+      "loss": 4.0176,
+      "step": 2565
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.869459450816942e-05,
+      "loss": 4.0331,
+      "step": 2570
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.86893273515421e-05,
+      "loss": 3.9665,
+      "step": 2575
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.868404973124139e-05,
+      "loss": 4.0069,
+      "step": 2580
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.86787616484015e-05,
+      "loss": 3.9786,
+      "step": 2585
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.867346310415886e-05,
+      "loss": 4.0387,
+      "step": 2590
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.866815409965219e-05,
+      "loss": 4.0319,
+      "step": 2595
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.86628346360224e-05,
+      "loss": 4.0779,
+      "step": 2600
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.865750471441271e-05,
+      "loss": 4.1236,
+      "step": 2605
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.865216433596853e-05,
+      "loss": 4.086,
+      "step": 2610
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.864681350183758e-05,
+      "loss": 4.0209,
+      "step": 2615
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.864145221316975e-05,
+      "loss": 3.8932,
+      "step": 2620
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.863608047111725e-05,
+      "loss": 4.1235,
+      "step": 2625
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.863069827683446e-05,
+      "loss": 4.023,
+      "step": 2630
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.862530563147812e-05,
+      "loss": 4.0363,
+      "step": 2635
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.861990253620709e-05,
+      "loss": 4.0139,
+      "step": 2640
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.861448899218255e-05,
+      "loss": 4.0695,
+      "step": 2645
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.860906500056792e-05,
+      "loss": 4.0567,
+      "step": 2650
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.860363056252883e-05,
+      "loss": 3.9513,
+      "step": 2655
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.859818567923321e-05,
+      "loss": 4.1195,
+      "step": 2660
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.859273035185118e-05,
+      "loss": 4.1449,
+      "step": 2665
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.858726458155512e-05,
+      "loss": 4.0953,
+      "step": 2670
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.858178836951967e-05,
+      "loss": 4.0136,
+      "step": 2675
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.857630171692174e-05,
+      "loss": 4.0779,
+      "step": 2680
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.857080462494039e-05,
+      "loss": 3.9711,
+      "step": 2685
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.856529709475703e-05,
+      "loss": 4.0123,
+      "step": 2690
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.855977912755524e-05,
+      "loss": 4.0099,
+      "step": 2695
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.855425072452086e-05,
+      "loss": 4.1222,
+      "step": 2700
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.854871188684204e-05,
+      "loss": 3.9998,
+      "step": 2705
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.854316261570905e-05,
+      "loss": 4.0798,
+      "step": 2710
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.85376029123145e-05,
+      "loss": 4.1759,
+      "step": 2715
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.853203277785318e-05,
+      "loss": 4.0731,
+      "step": 2720
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.85264522135222e-05,
+      "loss": 3.9181,
+      "step": 2725
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.852086122052083e-05,
+      "loss": 4.055,
+      "step": 2730
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.85152598000506e-05,
+      "loss": 4.0103,
+      "step": 2735
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.850964795331531e-05,
+      "loss": 3.9675,
+      "step": 2740
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.8504025681521e-05,
+      "loss": 4.1102,
+      "step": 2745
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.84983929858759e-05,
+      "loss": 3.9104,
+      "step": 2750
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.849274986759054e-05,
+      "loss": 3.9561,
+      "step": 2755
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.848709632787768e-05,
+      "loss": 4.055,
+      "step": 2760
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.848143236795225e-05,
+      "loss": 3.9903,
+      "step": 2765
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.847575798903152e-05,
+      "loss": 3.9108,
+      "step": 2770
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.847007319233496e-05,
+      "loss": 4.0207,
+      "step": 2775
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.846437797908424e-05,
+      "loss": 4.0715,
+      "step": 2780
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.845867235050333e-05,
+      "loss": 4.0179,
+      "step": 2785
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.845295630781837e-05,
+      "loss": 4.0214,
+      "step": 2790
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.844722985225783e-05,
+      "loss": 3.997,
+      "step": 2795
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.844149298505232e-05,
+      "loss": 3.99,
+      "step": 2800
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.843574570743474e-05,
+      "loss": 3.9435,
+      "step": 2805
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.842998802064025e-05,
+      "loss": 4.0337,
+      "step": 2810
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.842421992590619e-05,
+      "loss": 4.0059,
+      "step": 2815
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.841844142447215e-05,
+      "loss": 3.9924,
+      "step": 2820
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.841265251758e-05,
+      "loss": 3.9975,
+      "step": 2825
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.840685320647381e-05,
+      "loss": 4.0907,
+      "step": 2830
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.840104349239988e-05,
+      "loss": 3.9953,
+      "step": 2835
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.839522337660677e-05,
+      "loss": 3.9883,
+      "step": 2840
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.838939286034526e-05,
+      "loss": 4.0782,
+      "step": 2845
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.838355194486836e-05,
+      "loss": 3.9427,
+      "step": 2850
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.837770063143133e-05,
+      "loss": 3.9775,
+      "step": 2855
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.837183892129164e-05,
+      "loss": 4.0011,
+      "step": 2860
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.836596681570906e-05,
+      "loss": 3.943,
+      "step": 2865
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.83600843159455e-05,
+      "loss": 3.8867,
+      "step": 2870
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.835419142326517e-05,
+      "loss": 4.0097,
+      "step": 2875
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.834828813893448e-05,
+      "loss": 4.008,
+      "step": 2880
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.83423744642221e-05,
+      "loss": 4.0543,
+      "step": 2885
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.833645040039892e-05,
+      "loss": 4.0342,
+      "step": 2890
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.833051594873805e-05,
+      "loss": 3.9999,
+      "step": 2895
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.832457111051485e-05,
+      "loss": 3.9684,
+      "step": 2900
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.831861588700692e-05,
+      "loss": 3.9012,
+      "step": 2905
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.831265027949406e-05,
+      "loss": 3.9327,
+      "step": 2910
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.830667428925833e-05,
+      "loss": 3.8147,
+      "step": 2915
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.8300687917584e-05,
+      "loss": 4.034,
+      "step": 2920
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.82946911657576e-05,
+      "loss": 3.953,
+      "step": 2925
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.828868403506788e-05,
+      "loss": 4.0038,
+      "step": 2930
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.828266652680577e-05,
+      "loss": 3.9231,
+      "step": 2935
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.827663864226453e-05,
+      "loss": 3.9594,
+      "step": 2940
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.827060038273956e-05,
+      "loss": 3.9096,
+      "step": 2945
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.826455174952852e-05,
+      "loss": 3.8554,
+      "step": 2950
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.825849274393133e-05,
+      "loss": 3.9998,
+      "step": 2955
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.825242336725009e-05,
+      "loss": 4.0444,
+      "step": 2960
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.824634362078915e-05,
+      "loss": 3.9519,
+      "step": 2965
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.824025350585511e-05,
+      "loss": 3.8638,
+      "step": 2970
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.823415302375676e-05,
+      "loss": 3.9253,
+      "step": 2975
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.822804217580515e-05,
+      "loss": 3.9935,
+      "step": 2980
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.822192096331351e-05,
+      "loss": 3.8593,
+      "step": 2985
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.82157893875974e-05,
+      "loss": 3.9468,
+      "step": 2990
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.820964744997445e-05,
+      "loss": 3.9458,
+      "step": 2995
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.820349515176466e-05,
+      "loss": 3.9279,
+      "step": 3000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.819733249429019e-05,
+      "loss": 3.9525,
+      "step": 3005
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.819115947887545e-05,
+      "loss": 3.9607,
+      "step": 3010
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.818497610684705e-05,
+      "loss": 3.8035,
+      "step": 3015
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.817878237953383e-05,
+      "loss": 4.0535,
+      "step": 3020
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.817257829826689e-05,
+      "loss": 3.9239,
+      "step": 3025
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.81663638643795e-05,
+      "loss": 4.0873,
+      "step": 3030
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.81601390792072e-05,
+      "loss": 3.9612,
+      "step": 3035
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.815390394408776e-05,
+      "loss": 3.8893,
+      "step": 3040
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.814765846036111e-05,
+      "loss": 3.8967,
+      "step": 3045
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.81414026293695e-05,
+      "loss": 3.9423,
+      "step": 3050
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.813513645245729e-05,
+      "loss": 4.0063,
+      "step": 3055
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.812885993097118e-05,
+      "loss": 4.0111,
+      "step": 3060
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.812257306626002e-05,
+      "loss": 4.0138,
+      "step": 3065
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.811627585967487e-05,
+      "loss": 4.1024,
+      "step": 3070
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.810996831256909e-05,
+      "loss": 3.9115,
+      "step": 3075
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.81036504262982e-05,
+      "loss": 3.937,
+      "step": 3080
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.809732220221995e-05,
+      "loss": 4.0062,
+      "step": 3085
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.80909836416943e-05,
+      "loss": 3.9184,
+      "step": 3090
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.80846347460835e-05,
+      "loss": 3.9822,
+      "step": 3095
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.807827551675192e-05,
+      "loss": 3.885,
+      "step": 3100
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.807190595506623e-05,
+      "loss": 3.9799,
+      "step": 3105
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.80655260623953e-05,
+      "loss": 3.8721,
+      "step": 3110
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.805913584011019e-05,
+      "loss": 3.9281,
+      "step": 3115
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.805273528958423e-05,
+      "loss": 3.9125,
+      "step": 3120
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.804632441219293e-05,
+      "loss": 4.0076,
+      "step": 3125
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.803990320931402e-05,
+      "loss": 3.9736,
+      "step": 3130
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.803347168232746e-05,
+      "loss": 3.8803,
+      "step": 3135
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.802702983261545e-05,
+      "loss": 3.924,
+      "step": 3140
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.802057766156238e-05,
+      "loss": 4.0791,
+      "step": 3145
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.801411517055488e-05,
+      "loss": 3.9674,
+      "step": 3150
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.800764236098175e-05,
+      "loss": 3.9224,
+      "step": 3155
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.800115923423405e-05,
+      "loss": 3.9599,
+      "step": 3160
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.799466579170508e-05,
+      "loss": 3.7788,
+      "step": 3165
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.798816203479029e-05,
+      "loss": 3.9189,
+      "step": 3170
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.79816479648874e-05,
+      "loss": 3.9055,
+      "step": 3175
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.797512358339633e-05,
+      "loss": 3.9903,
+      "step": 3180
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.796858889171921e-05,
+      "loss": 4.016,
+      "step": 3185
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.79620438912604e-05,
+      "loss": 3.9585,
+      "step": 3190
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.795548858342645e-05,
+      "loss": 3.9805,
+      "step": 3195
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.794892296962615e-05,
+      "loss": 3.9203,
+      "step": 3200
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.794234705127047e-05,
+      "loss": 3.9033,
+      "step": 3205
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.793576082977267e-05,
+      "loss": 3.9066,
+      "step": 3210
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.792916430654815e-05,
+      "loss": 3.9179,
+      "step": 3215
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.792255748301453e-05,
+      "loss": 3.9567,
+      "step": 3220
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.791594036059168e-05,
+      "loss": 3.9365,
+      "step": 3225
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.790931294070169e-05,
+      "loss": 3.9199,
+      "step": 3230
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.79026752247688e-05,
+      "loss": 3.9036,
+      "step": 3235
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.78960272142195e-05,
+      "loss": 3.7834,
+      "step": 3240
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.788936891048251e-05,
+      "loss": 3.8953,
+      "step": 3245
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.788270031498876e-05,
+      "loss": 4.0192,
+      "step": 3250
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.787602142917135e-05,
+      "loss": 3.9296,
+      "step": 3255
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.786933225446563e-05,
+      "loss": 3.9252,
+      "step": 3260
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.786263279230915e-05,
+      "loss": 3.9483,
+      "step": 3265
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.785592304414167e-05,
+      "loss": 3.8679,
+      "step": 3270
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.784920301140516e-05,
+      "loss": 3.8496,
+      "step": 3275
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.78424726955438e-05,
+      "loss": 3.7942,
+      "step": 3280
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.783573209800398e-05,
+      "loss": 3.8962,
+      "step": 3285
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.782898122023432e-05,
+      "loss": 3.8798,
+      "step": 3290
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.78222200636856e-05,
+      "loss": 3.8346,
+      "step": 3295
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.781544862981087e-05,
+      "loss": 3.7781,
+      "step": 3300
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.780866692006534e-05,
+      "loss": 3.9586,
+      "step": 3305
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.780187493590645e-05,
+      "loss": 4.0283,
+      "step": 3310
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.779507267879385e-05,
+      "loss": 3.9209,
+      "step": 3315
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.778826015018936e-05,
+      "loss": 3.928,
+      "step": 3320
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.778143735155709e-05,
+      "loss": 3.8304,
+      "step": 3325
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.777460428436327e-05,
+      "loss": 3.9103,
+      "step": 3330
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.776776095007638e-05,
+      "loss": 3.896,
+      "step": 3335
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.776090735016711e-05,
+      "loss": 3.7634,
+      "step": 3340
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.775404348610835e-05,
+      "loss": 3.7548,
+      "step": 3345
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.774716935937518e-05,
+      "loss": 3.9337,
+      "step": 3350
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.77402849714449e-05,
+      "loss": 4.0026,
+      "step": 3355
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.773339032379701e-05,
+      "loss": 3.8779,
+      "step": 3360
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.772648541791323e-05,
+      "loss": 3.9515,
+      "step": 3365
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.771957025527745e-05,
+      "loss": 3.8295,
+      "step": 3370
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.771264483737581e-05,
+      "loss": 3.812,
+      "step": 3375
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.770570916569661e-05,
+      "loss": 3.8668,
+      "step": 3380
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.76987632417304e-05,
+      "loss": 3.8579,
+      "step": 3385
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.769180706696988e-05,
+      "loss": 3.9158,
+      "step": 3390
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.768484064290997e-05,
+      "loss": 3.7522,
+      "step": 3395
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.767786397104786e-05,
+      "loss": 3.8881,
+      "step": 3400
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.767087705288283e-05,
+      "loss": 4.0107,
+      "step": 3405
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.766387988991642e-05,
+      "loss": 3.9257,
+      "step": 3410
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.765687248365241e-05,
+      "loss": 3.9302,
+      "step": 3415
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.76498548355967e-05,
+      "loss": 3.8482,
+      "step": 3420
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.764282694725744e-05,
+      "loss": 3.8826,
+      "step": 3425
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.7635788820145e-05,
+      "loss": 3.829,
+      "step": 3430
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.762874045577189e-05,
+      "loss": 3.9172,
+      "step": 3435
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.762168185565284e-05,
+      "loss": 3.825,
+      "step": 3440
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.761461302130484e-05,
+      "loss": 3.9374,
+      "step": 3445
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.760753395424699e-05,
+      "loss": 3.8591,
+      "step": 3450
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.760044465600065e-05,
+      "loss": 3.7634,
+      "step": 3455
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.759334512808935e-05,
+      "loss": 3.8516,
+      "step": 3460
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.758623537203883e-05,
+      "loss": 3.8181,
+      "step": 3465
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.757911538937706e-05,
+      "loss": 3.7912,
+      "step": 3470
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.757198518163411e-05,
+      "loss": 3.8151,
+      "step": 3475
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.756484475034235e-05,
+      "loss": 3.8625,
+      "step": 3480
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.75576940970363e-05,
+      "loss": 3.9561,
+      "step": 3485
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.755053322325269e-05,
+      "loss": 3.816,
+      "step": 3490
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.754336213053043e-05,
+      "loss": 3.8605,
+      "step": 3495
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.753618082041066e-05,
+      "loss": 3.8019,
+      "step": 3500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.752898929443666e-05,
+      "loss": 3.9104,
+      "step": 3505
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.752178755415396e-05,
+      "loss": 3.815,
+      "step": 3510
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.751457560111026e-05,
+      "loss": 3.9031,
+      "step": 3515
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.750735343685547e-05,
+      "loss": 3.9328,
+      "step": 3520
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.750012106294167e-05,
+      "loss": 3.8333,
+      "step": 3525
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.749287848092315e-05,
+      "loss": 3.9392,
+      "step": 3530
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.748562569235637e-05,
+      "loss": 3.9369,
+      "step": 3535
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.747836269880003e-05,
+      "loss": 3.731,
+      "step": 3540
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.7471089501815e-05,
+      "loss": 3.9212,
+      "step": 3545
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.746380610296435e-05,
+      "loss": 3.769,
+      "step": 3550
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.74565125038133e-05,
+      "loss": 3.9181,
+      "step": 3555
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.744920870592931e-05,
+      "loss": 3.8753,
+      "step": 3560
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.7441894710882e-05,
+      "loss": 3.82,
+      "step": 3565
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.743457052024323e-05,
+      "loss": 3.9135,
+      "step": 3570
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.742723613558702e-05,
+      "loss": 3.8458,
+      "step": 3575
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.741989155848953e-05,
+      "loss": 3.7894,
+      "step": 3580
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.741253679052923e-05,
+      "loss": 3.796,
+      "step": 3585
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.740517183328666e-05,
+      "loss": 3.8105,
+      "step": 3590
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.739779668834461e-05,
+      "loss": 3.8756,
+      "step": 3595
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.739041135728806e-05,
+      "loss": 3.7961,
+      "step": 3600
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.738301584170417e-05,
+      "loss": 3.8702,
+      "step": 3605
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.737561014318228e-05,
+      "loss": 3.7876,
+      "step": 3610
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.736819426331392e-05,
+      "loss": 3.9309,
+      "step": 3615
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.736076820369283e-05,
+      "loss": 3.8683,
+      "step": 3620
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.735333196591492e-05,
+      "loss": 3.7648,
+      "step": 3625
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.734588555157826e-05,
+      "loss": 3.9061,
+      "step": 3630
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.733842896228318e-05,
+      "loss": 3.9367,
+      "step": 3635
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.733096219963211e-05,
+      "loss": 3.8217,
+      "step": 3640
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.732348526522974e-05,
+      "loss": 3.8459,
+      "step": 3645
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.73159981606829e-05,
+      "loss": 3.9581,
+      "step": 3650
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.730850088760064e-05,
+      "loss": 3.8337,
+      "step": 3655
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.730099344759413e-05,
+      "loss": 3.73,
+      "step": 3660
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.729347584227684e-05,
+      "loss": 3.8618,
+      "step": 3665
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.72859480732643e-05,
+      "loss": 3.8705,
+      "step": 3670
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.727841014217429e-05,
+      "loss": 3.6926,
+      "step": 3675
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.727086205062676e-05,
+      "loss": 3.9235,
+      "step": 3680
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.72633038002439e-05,
+      "loss": 3.8321,
+      "step": 3685
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.725573539264994e-05,
+      "loss": 3.816,
+      "step": 3690
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.724815682947146e-05,
+      "loss": 3.8725,
+      "step": 3695
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.72405681123371e-05,
+      "loss": 3.7806,
+      "step": 3700
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.723296924287775e-05,
+      "loss": 3.8389,
+      "step": 3705
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.722536022272646e-05,
+      "loss": 3.9748,
+      "step": 3710
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.721774105351845e-05,
+      "loss": 3.8451,
+      "step": 3715
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.721011173689114e-05,
+      "loss": 3.9031,
+      "step": 3720
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.720247227448411e-05,
+      "loss": 3.8523,
+      "step": 3725
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.719482266793916e-05,
+      "loss": 3.872,
+      "step": 3730
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.718716291890022e-05,
+      "loss": 3.7994,
+      "step": 3735
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.717949302901343e-05,
+      "loss": 3.809,
+      "step": 3740
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.71718129999271e-05,
+      "loss": 3.7294,
+      "step": 3745
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.716412283329172e-05,
+      "loss": 3.8161,
+      "step": 3750
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.715642253075998e-05,
+      "loss": 3.7014,
+      "step": 3755
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.714871209398671e-05,
+      "loss": 3.8454,
+      "step": 3760
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.714099152462892e-05,
+      "loss": 3.7692,
+      "step": 3765
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.713326082434584e-05,
+      "loss": 3.974,
+      "step": 3770
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.712551999479884e-05,
+      "loss": 3.7969,
+      "step": 3775
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.71177690376515e-05,
+      "loss": 3.8413,
+      "step": 3780
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.711000795456951e-05,
+      "loss": 3.8275,
+      "step": 3785
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.710223674722083e-05,
+      "loss": 3.8044,
+      "step": 3790
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.709445541727551e-05,
+      "loss": 3.8594,
+      "step": 3795
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.708666396640582e-05,
+      "loss": 3.8391,
+      "step": 3800
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.707886239628621e-05,
+      "loss": 3.7244,
+      "step": 3805
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.707105070859326e-05,
+      "loss": 3.8842,
+      "step": 3810
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.706322890500579e-05,
+      "loss": 3.8671,
+      "step": 3815
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.705539698720476e-05,
+      "loss": 3.8995,
+      "step": 3820
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.704755495687328e-05,
+      "loss": 3.7689,
+      "step": 3825
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.703970281569668e-05,
+      "loss": 3.781,
+      "step": 3830
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.70318405653624e-05,
+      "loss": 3.8412,
+      "step": 3835
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.702396820756014e-05,
+      "loss": 3.8823,
+      "step": 3840
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.701608574398169e-05,
+      "loss": 3.9033,
+      "step": 3845
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.700819317632107e-05,
+      "loss": 3.847,
+      "step": 3850
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.700029050627443e-05,
+      "loss": 3.7976,
+      "step": 3855
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.699237773554012e-05,
+      "loss": 3.8756,
+      "step": 3860
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.698445486581864e-05,
+      "loss": 3.8885,
+      "step": 3865
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.697652189881266e-05,
+      "loss": 3.7132,
+      "step": 3870
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.696857883622706e-05,
+      "loss": 3.8192,
+      "step": 3875
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.696062567976885e-05,
+      "loss": 3.8404,
+      "step": 3880
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.695266243114718e-05,
+      "loss": 3.8806,
+      "step": 3885
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.694468909207345e-05,
+      "loss": 3.7846,
+      "step": 3890
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.693670566426119e-05,
+      "loss": 3.784,
+      "step": 3895
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.692871214942608e-05,
+      "loss": 3.8687,
+      "step": 3900
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.692070854928596e-05,
+      "loss": 3.846,
+      "step": 3905
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.691269486556088e-05,
+      "loss": 3.7714,
+      "step": 3910
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.690467109997303e-05,
+      "loss": 3.8154,
+      "step": 3915
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.68966372542468e-05,
+      "loss": 3.8492,
+      "step": 3920
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.688859333010866e-05,
+      "loss": 3.6749,
+      "step": 3925
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.688053932928737e-05,
+      "loss": 3.7613,
+      "step": 3930
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.687247525351375e-05,
+      "loss": 3.7665,
+      "step": 3935
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.686440110452084e-05,
+      "loss": 3.8007,
+      "step": 3940
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.685631688404384e-05,
+      "loss": 3.7874,
+      "step": 3945
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.684822259382008e-05,
+      "loss": 3.8367,
+      "step": 3950
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.684011823558909e-05,
+      "loss": 3.9224,
+      "step": 3955
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.683200381109254e-05,
+      "loss": 3.7707,
+      "step": 3960
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.68238793220743e-05,
+      "loss": 3.7687,
+      "step": 3965
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.681574477028039e-05,
+      "loss": 3.777,
+      "step": 3970
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.680760015745894e-05,
+      "loss": 3.8291,
+      "step": 3975
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.67994454853603e-05,
+      "loss": 3.8183,
+      "step": 3980
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.679128075573698e-05,
+      "loss": 3.7526,
+      "step": 3985
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.678310597034364e-05,
+      "loss": 3.8308,
+      "step": 3990
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.677492113093706e-05,
+      "loss": 3.7682,
+      "step": 3995
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.676672623927625e-05,
+      "loss": 3.7863,
+      "step": 4000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.675852129712234e-05,
+      "loss": 3.8452,
+      "step": 4005
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.675030630623864e-05,
+      "loss": 3.8182,
+      "step": 4010
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.67420812683906e-05,
+      "loss": 3.8417,
+      "step": 4015
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.673384618534582e-05,
+      "loss": 3.7864,
+      "step": 4020
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.672560105887409e-05,
+      "loss": 3.6989,
+      "step": 4025
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.671734589074736e-05,
+      "loss": 3.8849,
+      "step": 4030
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.670908068273972e-05,
+      "loss": 3.9271,
+      "step": 4035
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.67008054366274e-05,
+      "loss": 3.7933,
+      "step": 4040
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.669252015418882e-05,
+      "loss": 3.7739,
+      "step": 4045
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.668422483720456e-05,
+      "loss": 3.7557,
+      "step": 4050
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.66759194874573e-05,
+      "loss": 3.7863,
+      "step": 4055
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.666760410673195e-05,
+      "loss": 3.7767,
+      "step": 4060
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.665927869681553e-05,
+      "loss": 3.8872,
+      "step": 4065
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.665094325949723e-05,
+      "loss": 3.866,
+      "step": 4070
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.664259779656842e-05,
+      "loss": 3.8484,
+      "step": 4075
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.663424230982254e-05,
+      "loss": 3.8315,
+      "step": 4080
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.66258768010553e-05,
+      "loss": 3.8183,
+      "step": 4085
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.661750127206445e-05,
+      "loss": 3.6897,
+      "step": 4090
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.660911572465e-05,
+      "loss": 3.8456,
+      "step": 4095
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.660072016061406e-05,
+      "loss": 3.7448,
+      "step": 4100
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.659231458176085e-05,
+      "loss": 3.7249,
+      "step": 4105
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.658389898989684e-05,
+      "loss": 3.8034,
+      "step": 4110
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.657547338683055e-05,
+      "loss": 3.8515,
+      "step": 4115
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.656703777437273e-05,
+      "loss": 3.6943,
+      "step": 4120
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.655859215433625e-05,
+      "loss": 3.6515,
+      "step": 4125
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.655013652853612e-05,
+      "loss": 3.6814,
+      "step": 4130
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.654167089878952e-05,
+      "loss": 3.869,
+      "step": 4135
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.653319526691577e-05,
+      "loss": 3.7586,
+      "step": 4140
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.652470963473635e-05,
+      "loss": 3.8997,
+      "step": 4145
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.651621400407487e-05,
+      "loss": 3.6785,
+      "step": 4150
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.650770837675712e-05,
+      "loss": 3.8194,
+      "step": 4155
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.649919275461099e-05,
+      "loss": 3.7334,
+      "step": 4160
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.649066713946655e-05,
+      "loss": 3.7594,
+      "step": 4165
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.648213153315603e-05,
+      "loss": 3.7695,
+      "step": 4170
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.647358593751378e-05,
+      "loss": 3.8279,
+      "step": 4175
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.646503035437633e-05,
+      "loss": 3.6245,
+      "step": 4180
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.64564647855823e-05,
+      "loss": 3.7828,
+      "step": 4185
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.644788923297253e-05,
+      "loss": 3.8807,
+      "step": 4190
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.64393036983899e-05,
+      "loss": 3.7854,
+      "step": 4195
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.643070818367957e-05,
+      "loss": 3.8569,
+      "step": 4200
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.642210269068874e-05,
+      "loss": 3.7832,
+      "step": 4205
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.641348722126679e-05,
+      "loss": 3.8536,
+      "step": 4210
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.640486177726527e-05,
+      "loss": 3.7589,
+      "step": 4215
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.639622636053781e-05,
+      "loss": 3.7834,
+      "step": 4220
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.638758097294024e-05,
+      "loss": 3.7669,
+      "step": 4225
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.637892561633054e-05,
+      "loss": 3.7446,
+      "step": 4230
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.637026029256874e-05,
+      "loss": 3.7606,
+      "step": 4235
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.636158500351713e-05,
+      "loss": 3.7689,
+      "step": 4240
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.635289975104009e-05,
+      "loss": 3.8526,
+      "step": 4245
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.634420453700413e-05,
+      "loss": 3.7471,
+      "step": 4250
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.633549936327789e-05,
+      "loss": 3.7281,
+      "step": 4255
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.63267842317322e-05,
+      "loss": 3.723,
+      "step": 4260
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.631805914424e-05,
+      "loss": 3.7395,
+      "step": 4265
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.630932410267637e-05,
+      "loss": 3.7371,
+      "step": 4270
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.630057910891852e-05,
+      "loss": 3.7576,
+      "step": 4275
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.629182416484582e-05,
+      "loss": 3.7184,
+      "step": 4280
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.628305927233975e-05,
+      "loss": 3.8061,
+      "step": 4285
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.627428443328397e-05,
+      "loss": 3.7949,
+      "step": 4290
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.626549964956424e-05,
+      "loss": 3.7399,
+      "step": 4295
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.625670492306848e-05,
+      "loss": 3.6738,
+      "step": 4300
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.624790025568673e-05,
+      "loss": 3.8158,
+      "step": 4305
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.623908564931117e-05,
+      "loss": 3.8368,
+      "step": 4310
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.623026110583612e-05,
+      "loss": 3.7656,
+      "step": 4315
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.622142662715808e-05,
+      "loss": 3.7445,
+      "step": 4320
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.621258221517556e-05,
+      "loss": 3.84,
+      "step": 4325
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.620372787178934e-05,
+      "loss": 3.8206,
+      "step": 4330
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.619486359890228e-05,
+      "loss": 3.7572,
+      "step": 4335
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.618598939841934e-05,
+      "loss": 3.8502,
+      "step": 4340
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.617710527224768e-05,
+      "loss": 3.8059,
+      "step": 4345
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.616821122229656e-05,
+      "loss": 3.668,
+      "step": 4350
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.615930725047732e-05,
+      "loss": 3.692,
+      "step": 4355
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.615039335870357e-05,
+      "loss": 3.8211,
+      "step": 4360
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.614146954889091e-05,
+      "loss": 3.8037,
+      "step": 4365
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.613253582295714e-05,
+      "loss": 3.7287,
+      "step": 4370
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.612359218282216e-05,
+      "loss": 3.8129,
+      "step": 4375
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.611463863040807e-05,
+      "loss": 3.7402,
+      "step": 4380
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.610567516763902e-05,
+      "loss": 3.773,
+      "step": 4385
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.60967017964413e-05,
+      "loss": 3.782,
+      "step": 4390
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.608771851874338e-05,
+      "loss": 3.7435,
+      "step": 4395
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.607872533647584e-05,
+      "loss": 3.7772,
+      "step": 4400
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.606972225157134e-05,
+      "loss": 3.7345,
+      "step": 4405
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.606070926596471e-05,
+      "loss": 3.8394,
+      "step": 4410
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.605168638159294e-05,
+      "loss": 3.7974,
+      "step": 4415
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.604265360039507e-05,
+      "loss": 3.6399,
+      "step": 4420
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.603361092431234e-05,
+      "loss": 3.7509,
+      "step": 4425
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.602455835528804e-05,
+      "loss": 3.727,
+      "step": 4430
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.601549589526767e-05,
+      "loss": 3.7592,
+      "step": 4435
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.60064235461988e-05,
+      "loss": 3.7137,
+      "step": 4440
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.599734131003114e-05,
+      "loss": 3.7122,
+      "step": 4445
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.598824918871654e-05,
+      "loss": 3.7267,
+      "step": 4450
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.597914718420892e-05,
+      "loss": 3.7377,
+      "step": 4455
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.597003529846442e-05,
+      "loss": 3.7321,
+      "step": 4460
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.59609135334412e-05,
+      "loss": 3.7409,
+      "step": 4465
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.595178189109963e-05,
+      "loss": 3.6241,
+      "step": 4470
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.594264037340212e-05,
+      "loss": 3.7963,
+      "step": 4475
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.593348898231326e-05,
+      "loss": 3.6321,
+      "step": 4480
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.592432771979978e-05,
+      "loss": 3.6509,
+      "step": 4485
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.591515658783046e-05,
+      "loss": 3.7897,
+      "step": 4490
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.590597558837626e-05,
+      "loss": 3.7453,
+      "step": 4495
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.589678472341023e-05,
+      "loss": 3.6429,
+      "step": 4500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.588758399490755e-05,
+      "loss": 3.8,
+      "step": 4505
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.587837340484553e-05,
+      "loss": 3.7181,
+      "step": 4510
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.586915295520358e-05,
+      "loss": 3.7177,
+      "step": 4515
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.585992264796324e-05,
+      "loss": 3.6759,
+      "step": 4520
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.585068248510817e-05,
+      "loss": 3.7835,
+      "step": 4525
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.584143246862414e-05,
+      "loss": 3.6714,
+      "step": 4530
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.583217260049907e-05,
+      "loss": 3.7505,
+      "step": 4535
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.582290288272291e-05,
+      "loss": 3.7962,
+      "step": 4540
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.581362331728784e-05,
+      "loss": 3.7605,
+      "step": 4545
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.580433390618808e-05,
+      "loss": 3.7017,
+      "step": 4550
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.579503465142e-05,
+      "loss": 3.7899,
+      "step": 4555
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.578572555498207e-05,
+      "loss": 3.6949,
+      "step": 4560
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.577640661887488e-05,
+      "loss": 3.7822,
+      "step": 4565
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.576707784510113e-05,
+      "loss": 3.7491,
+      "step": 4570
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.575773923566563e-05,
+      "loss": 3.7906,
+      "step": 4575
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.574839079257532e-05,
+      "loss": 3.6757,
+      "step": 4580
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.573903251783925e-05,
+      "loss": 3.7124,
+      "step": 4585
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.572966441346858e-05,
+      "loss": 3.6875,
+      "step": 4590
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.572028648147657e-05,
+      "loss": 3.7993,
+      "step": 4595
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.571089872387862e-05,
+      "loss": 3.7606,
+      "step": 4600
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.57015011426922e-05,
+      "loss": 3.7155,
+      "step": 4605
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.569209373993693e-05,
+      "loss": 3.7319,
+      "step": 4610
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.568267651763451e-05,
+      "loss": 3.8139,
+      "step": 4615
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.56732494778088e-05,
+      "loss": 3.689,
+      "step": 4620
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.56638126224857e-05,
+      "loss": 3.5653,
+      "step": 4625
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.565436595369328e-05,
+      "loss": 3.6938,
+      "step": 4630
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.564490947346168e-05,
+      "loss": 3.7769,
+      "step": 4635
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.563544318382316e-05,
+      "loss": 3.7478,
+      "step": 4640
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.56259670868121e-05,
+      "loss": 3.7228,
+      "step": 4645
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.5616481184465e-05,
+      "loss": 3.7254,
+      "step": 4650
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.560698547882041e-05,
+      "loss": 3.7988,
+      "step": 4655
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.559747997191903e-05,
+      "loss": 3.7504,
+      "step": 4660
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.558796466580366e-05,
+      "loss": 3.7322,
+      "step": 4665
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.557843956251924e-05,
+      "loss": 3.6156,
+      "step": 4670
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.556890466411272e-05,
+      "loss": 3.6691,
+      "step": 4675
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.555935997263328e-05,
+      "loss": 3.7377,
+      "step": 4680
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.554980549013209e-05,
+      "loss": 3.6093,
+      "step": 4685
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.554024121866249e-05,
+      "loss": 3.732,
+      "step": 4690
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.55306671602799e-05,
+      "loss": 3.8209,
+      "step": 4695
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.552108331704189e-05,
+      "loss": 3.7286,
+      "step": 4700
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.551148969100804e-05,
+      "loss": 3.7973,
+      "step": 4705
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.550188628424012e-05,
+      "loss": 3.7771,
+      "step": 4710
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.549227309880197e-05,
+      "loss": 3.7171,
+      "step": 4715
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.548265013675951e-05,
+      "loss": 3.6423,
+      "step": 4720
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.54730174001808e-05,
+      "loss": 3.7968,
+      "step": 4725
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.546337489113597e-05,
+      "loss": 3.8109,
+      "step": 4730
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.545372261169725e-05,
+      "loss": 3.755,
+      "step": 4735
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.544406056393903e-05,
+      "loss": 3.6482,
+      "step": 4740
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.54343887499377e-05,
+      "loss": 3.7506,
+      "step": 4745
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.542470717177182e-05,
+      "loss": 3.7,
+      "step": 4750
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.541501583152204e-05,
+      "loss": 3.551,
+      "step": 4755
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.540531473127106e-05,
+      "loss": 3.6974,
+      "step": 4760
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.539560387310375e-05,
+      "loss": 3.7426,
+      "step": 4765
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.538588325910702e-05,
+      "loss": 3.7302,
+      "step": 4770
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.53761528913699e-05,
+      "loss": 3.6671,
+      "step": 4775
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.536641277198353e-05,
+      "loss": 3.6741,
+      "step": 4780
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.535666290304111e-05,
+      "loss": 3.6784,
+      "step": 4785
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.534690328663796e-05,
+      "loss": 3.6839,
+      "step": 4790
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.533713392487149e-05,
+      "loss": 3.7798,
+      "step": 4795
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.532735481984121e-05,
+      "loss": 3.711,
+      "step": 4800
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.53175659736487e-05,
+      "loss": 3.7949,
+      "step": 4805
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.530776738839766e-05,
+      "loss": 3.809,
+      "step": 4810
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.529795906619387e-05,
+      "loss": 3.7112,
+      "step": 4815
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.528814100914522e-05,
+      "loss": 3.6319,
+      "step": 4820
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.52783132193617e-05,
+      "loss": 3.7483,
+      "step": 4825
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.52684756989553e-05,
+      "loss": 3.7475,
+      "step": 4830
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.525862845004022e-05,
+      "loss": 3.6521,
+      "step": 4835
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.52487714747327e-05,
+      "loss": 3.708,
+      "step": 4840
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.523890477515105e-05,
+      "loss": 3.6546,
+      "step": 4845
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.522902835341573e-05,
+      "loss": 3.7265,
+      "step": 4850
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.521914221164922e-05,
+      "loss": 3.7473,
+      "step": 4855
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.520924635197612e-05,
+      "loss": 3.7208,
+      "step": 4860
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.519934077652312e-05,
+      "loss": 3.7653,
+      "step": 4865
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.518942548741902e-05,
+      "loss": 3.6918,
+      "step": 4870
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.517950048679467e-05,
+      "loss": 3.6694,
+      "step": 4875
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.516956577678302e-05,
+      "loss": 3.714,
+      "step": 4880
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.51596213595191e-05,
+      "loss": 3.7167,
+      "step": 4885
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.514966723714004e-05,
+      "loss": 3.7339,
+      "step": 4890
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.513970341178506e-05,
+      "loss": 3.6999,
+      "step": 4895
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.512972988559545e-05,
+      "loss": 3.6182,
+      "step": 4900
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.511974666071458e-05,
+      "loss": 3.5643,
+      "step": 4905
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.510975373928792e-05,
+      "loss": 3.7679,
+      "step": 4910
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.509975112346303e-05,
+      "loss": 3.7247,
+      "step": 4915
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.508973881538952e-05,
+      "loss": 3.7034,
+      "step": 4920
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.507971681721911e-05,
+      "loss": 3.6121,
+      "step": 4925
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.50696851311056e-05,
+      "loss": 3.7025,
+      "step": 4930
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.505964375920489e-05,
+      "loss": 3.6147,
+      "step": 4935
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.50495927036749e-05,
+      "loss": 3.7171,
+      "step": 4940
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.503953196667569e-05,
+      "loss": 3.7066,
+      "step": 4945
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.502946155036938e-05,
+      "loss": 3.6834,
+      "step": 4950
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.50193814569202e-05,
+      "loss": 3.6803,
+      "step": 4955
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.500929168849439e-05,
+      "loss": 3.6684,
+      "step": 4960
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.499919224726032e-05,
+      "loss": 3.7299,
+      "step": 4965
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.498908313538846e-05,
+      "loss": 3.5976,
+      "step": 4970
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.49789643550513e-05,
+      "loss": 3.5527,
+      "step": 4975
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.496883590842343e-05,
+      "loss": 3.7427,
+      "step": 4980
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.495869779768154e-05,
+      "loss": 3.6779,
+      "step": 4985
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.494855002500438e-05,
+      "loss": 3.736,
+      "step": 4990
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.493839259257278e-05,
+      "loss": 3.7085,
+      "step": 4995
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.492822550256962e-05,
+      "loss": 3.627,
+      "step": 5000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.49180487571799e-05,
+      "loss": 3.7048,
+      "step": 5005
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.490786235859066e-05,
+      "loss": 3.7442,
+      "step": 5010
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.489766630899104e-05,
+      "loss": 3.6095,
+      "step": 5015
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.488746061057223e-05,
+      "loss": 3.7268,
+      "step": 5020
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.487724526552753e-05,
+      "loss": 3.7492,
+      "step": 5025
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.486702027605224e-05,
+      "loss": 3.6174,
+      "step": 5030
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.485678564434381e-05,
+      "loss": 3.7855,
+      "step": 5035
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.484654137260176e-05,
+      "loss": 3.6784,
+      "step": 5040
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.483628746302761e-05,
+      "loss": 3.6298,
+      "step": 5045
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.482602391782501e-05,
+      "loss": 3.6547,
+      "step": 5050
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.481575073919968e-05,
+      "loss": 3.6159,
+      "step": 5055
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.480546792935939e-05,
+      "loss": 3.7578,
+      "step": 5060
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.479517549051398e-05,
+      "loss": 3.7054,
+      "step": 5065
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.478487342487536e-05,
+      "loss": 3.7131,
+      "step": 5070
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.477456173465754e-05,
+      "loss": 3.7345,
+      "step": 5075
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.476424042207655e-05,
+      "loss": 3.6158,
+      "step": 5080
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.475390948935054e-05,
+      "loss": 3.5713,
+      "step": 5085
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.474356893869967e-05,
+      "loss": 3.7855,
+      "step": 5090
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.47332187723462e-05,
+      "loss": 3.6766,
+      "step": 5095
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.472285899251445e-05,
+      "loss": 3.695,
+      "step": 5100
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.471248960143085e-05,
+      "loss": 3.6845,
+      "step": 5105
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.47021106013238e-05,
+      "loss": 3.6886,
+      "step": 5110
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.469172199442383e-05,
+      "loss": 3.6684,
+      "step": 5115
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.468132378296355e-05,
+      "loss": 3.7259,
+      "step": 5120
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.467091596917758e-05,
+      "loss": 3.6495,
+      "step": 5125
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.466049855530265e-05,
+      "loss": 3.649,
+      "step": 5130
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.465007154357752e-05,
+      "loss": 3.8183,
+      "step": 5135
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.463963493624304e-05,
+      "loss": 3.6724,
+      "step": 5140
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.462918873554209e-05,
+      "loss": 3.7299,
+      "step": 5145
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.461873294371965e-05,
+      "loss": 3.6499,
+      "step": 5150
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.460826756302275e-05,
+      "loss": 3.6142,
+      "step": 5155
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.459779259570045e-05,
+      "loss": 3.6762,
+      "step": 5160
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.458730804400389e-05,
+      "loss": 3.6228,
+      "step": 5165
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.457681391018629e-05,
+      "loss": 3.7529,
+      "step": 5170
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.45663101965029e-05,
+      "loss": 3.7099,
+      "step": 5175
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.455579690521106e-05,
+      "loss": 3.5869,
+      "step": 5180
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.454527403857013e-05,
+      "loss": 3.7428,
+      "step": 5185
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.453474159884155e-05,
+      "loss": 3.6858,
+      "step": 5190
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.452419958828883e-05,
+      "loss": 3.7065,
+      "step": 5195
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.45136480091775e-05,
+      "loss": 3.7555,
+      "step": 5200
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.450308686377516e-05,
+      "loss": 3.7485,
+      "step": 5205
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.449251615435151e-05,
+      "loss": 3.6872,
+      "step": 5210
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.448193588317822e-05,
+      "loss": 3.7424,
+      "step": 5215
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.447134605252912e-05,
+      "loss": 3.6962,
+      "step": 5220
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.446074666468e-05,
+      "loss": 3.6245,
+      "step": 5225
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.445013772190874e-05,
+      "loss": 3.6098,
+      "step": 5230
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.443951922649529e-05,
+      "loss": 3.6804,
+      "step": 5235
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.442889118072162e-05,
+      "loss": 3.7355,
+      "step": 5240
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.441825358687181e-05,
+      "loss": 3.7511,
+      "step": 5245
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.440760644723192e-05,
+      "loss": 3.7018,
+      "step": 5250
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.439694976409008e-05,
+      "loss": 3.5722,
+      "step": 5255
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.438628353973653e-05,
+      "loss": 3.681,
+      "step": 5260
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.437560777646348e-05,
+      "loss": 3.7316,
+      "step": 5265
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.436492247656523e-05,
+      "loss": 3.6133,
+      "step": 5270
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.435422764233814e-05,
+      "loss": 3.8114,
+      "step": 5275
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.43435232760806e-05,
+      "loss": 3.6959,
+      "step": 5280
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.433280938009304e-05,
+      "loss": 3.6721,
+      "step": 5285
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.432208595667797e-05,
+      "loss": 3.6609,
+      "step": 5290
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.431135300813989e-05,
+      "loss": 3.7046,
+      "step": 5295
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.430061053678542e-05,
+      "loss": 3.6872,
+      "step": 5300
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.42898585449232e-05,
+      "loss": 3.6058,
+      "step": 5305
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.427909703486388e-05,
+      "loss": 3.669,
+      "step": 5310
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.426832600892019e-05,
+      "loss": 3.6405,
+      "step": 5315
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.425754546940691e-05,
+      "loss": 3.648,
+      "step": 5320
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.424675541864081e-05,
+      "loss": 3.6683,
+      "step": 5325
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.423595585894082e-05,
+      "loss": 3.6513,
+      "step": 5330
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.422514679262778e-05,
+      "loss": 3.6118,
+      "step": 5335
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.421432822202465e-05,
+      "loss": 3.7514,
+      "step": 5340
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.420350014945643e-05,
+      "loss": 3.6282,
+      "step": 5345
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.419266257725013e-05,
+      "loss": 3.6855,
+      "step": 5350
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.418181550773483e-05,
+      "loss": 3.6075,
+      "step": 5355
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.417095894324163e-05,
+      "loss": 3.683,
+      "step": 5360
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.41600928861037e-05,
+      "loss": 3.6887,
+      "step": 5365
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.41492173386562e-05,
+      "loss": 3.7112,
+      "step": 5370
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.413833230323639e-05,
+      "loss": 3.6634,
+      "step": 5375
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.412743778218351e-05,
+      "loss": 3.6871,
+      "step": 5380
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.41165337778389e-05,
+      "loss": 3.6577,
+      "step": 5385
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.410562029254589e-05,
+      "loss": 3.6844,
+      "step": 5390
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.409469732864986e-05,
+      "loss": 3.6785,
+      "step": 5395
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.408376488849824e-05,
+      "loss": 3.5706,
+      "step": 5400
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.407282297444048e-05,
+      "loss": 3.5663,
+      "step": 5405
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.406187158882808e-05,
+      "loss": 3.6937,
+      "step": 5410
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.405091073401458e-05,
+      "loss": 3.7414,
+      "step": 5415
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.40399404123555e-05,
+      "loss": 3.5024,
+      "step": 5420
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.402896062620848e-05,
+      "loss": 3.6385,
+      "step": 5425
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.401797137793313e-05,
+      "loss": 3.6656,
+      "step": 5430
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.400697266989112e-05,
+      "loss": 3.651,
+      "step": 5435
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.399596450444617e-05,
+      "loss": 3.6819,
+      "step": 5440
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.398494688396398e-05,
+      "loss": 3.6068,
+      "step": 5445
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.397391981081232e-05,
+      "loss": 3.5839,
+      "step": 5450
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.3962883287361e-05,
+      "loss": 3.6475,
+      "step": 5455
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.395183731598183e-05,
+      "loss": 3.6129,
+      "step": 5460
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.394078189904867e-05,
+      "loss": 3.5024,
+      "step": 5465
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.392971703893741e-05,
+      "loss": 3.536,
+      "step": 5470
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.391864273802595e-05,
+      "loss": 3.6793,
+      "step": 5475
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.390755899869425e-05,
+      "loss": 3.662,
+      "step": 5480
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.38964658233243e-05,
+      "loss": 3.6093,
+      "step": 5485
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.388536321430005e-05,
+      "loss": 3.7504,
+      "step": 5490
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.387425117400757e-05,
+      "loss": 3.552,
+      "step": 5495
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.386312970483489e-05,
+      "loss": 3.5696,
+      "step": 5500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.385199880917211e-05,
+      "loss": 3.6238,
+      "step": 5505
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.384085848941132e-05,
+      "loss": 3.6879,
+      "step": 5510
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.382970874794666e-05,
+      "loss": 3.641,
+      "step": 5515
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.381854958717429e-05,
+      "loss": 3.5246,
+      "step": 5520
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.380738100949238e-05,
+      "loss": 3.6289,
+      "step": 5525
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.379620301730114e-05,
+      "loss": 3.5976,
+      "step": 5530
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.378501561300282e-05,
+      "loss": 3.5259,
+      "step": 5535
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.377381879900162e-05,
+      "loss": 3.5889,
+      "step": 5540
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.376261257770384e-05,
+      "loss": 3.5694,
+      "step": 5545
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.37513969515178e-05,
+      "loss": 3.6377,
+      "step": 5550
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.374017192285378e-05,
+      "loss": 3.5271,
+      "step": 5555
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.372893749412413e-05,
+      "loss": 3.6465,
+      "step": 5560
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.371769366774321e-05,
+      "loss": 3.5937,
+      "step": 5565
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.370644044612738e-05,
+      "loss": 3.7053,
+      "step": 5570
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.369517783169505e-05,
+      "loss": 3.5916,
+      "step": 5575
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.368390582686663e-05,
+      "loss": 3.6582,
+      "step": 5580
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.367262443406456e-05,
+      "loss": 3.6273,
+      "step": 5585
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.366133365571328e-05,
+      "loss": 3.7097,
+      "step": 5590
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.365003349423924e-05,
+      "loss": 3.6454,
+      "step": 5595
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.363872395207097e-05,
+      "loss": 3.5933,
+      "step": 5600
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.362740503163891e-05,
+      "loss": 3.735,
+      "step": 5605
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.361607673537563e-05,
+      "loss": 3.6228,
+      "step": 5610
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.36047390657156e-05,
+      "loss": 3.6316,
+      "step": 5615
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.359339202509543e-05,
+      "loss": 3.6303,
+      "step": 5620
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.358203561595363e-05,
+      "loss": 3.6156,
+      "step": 5625
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.357066984073079e-05,
+      "loss": 3.6543,
+      "step": 5630
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.355929470186948e-05,
+      "loss": 3.4899,
+      "step": 5635
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.354791020181431e-05,
+      "loss": 3.7219,
+      "step": 5640
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.353651634301188e-05,
+      "loss": 3.5609,
+      "step": 5645
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.352511312791083e-05,
+      "loss": 3.5893,
+      "step": 5650
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.351370055896175e-05,
+      "loss": 3.5975,
+      "step": 5655
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.350227863861732e-05,
+      "loss": 3.6707,
+      "step": 5660
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.349084736933215e-05,
+      "loss": 3.6515,
+      "step": 5665
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.347940675356292e-05,
+      "loss": 3.6141,
+      "step": 5670
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.346795679376832e-05,
+      "loss": 3.5988,
+      "step": 5675
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.345649749240898e-05,
+      "loss": 3.6056,
+      "step": 5680
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.344502885194763e-05,
+      "loss": 3.5654,
+      "step": 5685
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.343355087484894e-05,
+      "loss": 3.5865,
+      "step": 5690
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.342206356357956e-05,
+      "loss": 3.5723,
+      "step": 5695
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.341056692060828e-05,
+      "loss": 3.5729,
+      "step": 5700
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.339906094840575e-05,
+      "loss": 3.6581,
+      "step": 5705
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.33875456494447e-05,
+      "loss": 3.6914,
+      "step": 5710
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.337602102619985e-05,
+      "loss": 3.6459,
+      "step": 5715
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.336448708114789e-05,
+      "loss": 3.6846,
+      "step": 5720
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.33529438167676e-05,
+      "loss": 3.6587,
+      "step": 5725
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.334139123553967e-05,
+      "loss": 3.6845,
+      "step": 5730
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.332982933994683e-05,
+      "loss": 3.4873,
+      "step": 5735
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.331825813247381e-05,
+      "loss": 3.6119,
+      "step": 5740
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.330667761560739e-05,
+      "loss": 3.6167,
+      "step": 5745
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.329508779183623e-05,
+      "loss": 3.7076,
+      "step": 5750
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.328348866365111e-05,
+      "loss": 3.5744,
+      "step": 5755
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.327188023354474e-05,
+      "loss": 3.7051,
+      "step": 5760
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.326026250401188e-05,
+      "loss": 3.5623,
+      "step": 5765
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.324863547754924e-05,
+      "loss": 3.651,
+      "step": 5770
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.323699915665555e-05,
+      "loss": 3.4928,
+      "step": 5775
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.322535354383153e-05,
+      "loss": 3.6745,
+      "step": 5780
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.321369864157992e-05,
+      "loss": 3.6986,
+      "step": 5785
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.320203445240543e-05,
+      "loss": 3.6629,
+      "step": 5790
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.319036097881477e-05,
+      "loss": 3.6185,
+      "step": 5795
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.317867822331667e-05,
+      "loss": 3.578,
+      "step": 5800
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.31669861884218e-05,
+      "loss": 3.6336,
+      "step": 5805
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.315528487664289e-05,
+      "loss": 3.7053,
+      "step": 5810
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.314357429049462e-05,
+      "loss": 3.5963,
+      "step": 5815
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.313185443249367e-05,
+      "loss": 3.6103,
+      "step": 5820
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.312012530515875e-05,
+      "loss": 3.6239,
+      "step": 5825
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.310838691101048e-05,
+      "loss": 3.6486,
+      "step": 5830
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.309663925257156e-05,
+      "loss": 3.6725,
+      "step": 5835
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.308488233236663e-05,
+      "loss": 3.6615,
+      "step": 5840
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.307311615292234e-05,
+      "loss": 3.636,
+      "step": 5845
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.306134071676734e-05,
+      "loss": 3.6853,
+      "step": 5850
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.30495560264322e-05,
+      "loss": 3.6074,
+      "step": 5855
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.303776208444958e-05,
+      "loss": 3.5875,
+      "step": 5860
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.302595889335405e-05,
+      "loss": 3.6085,
+      "step": 5865
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.301414645568223e-05,
+      "loss": 3.675,
+      "step": 5870
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.300232477397266e-05,
+      "loss": 3.5822,
+      "step": 5875
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.299049385076592e-05,
+      "loss": 3.6136,
+      "step": 5880
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.297865368860456e-05,
+      "loss": 3.5517,
+      "step": 5885
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.296680429003311e-05,
+      "loss": 3.6539,
+      "step": 5890
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.295494565759808e-05,
+      "loss": 3.4057,
+      "step": 5895
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.294307779384796e-05,
+      "loss": 3.4734,
+      "step": 5900
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.293120070133327e-05,
+      "loss": 3.588,
+      "step": 5905
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.291931438260646e-05,
+      "loss": 3.5634,
+      "step": 5910
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.290741884022198e-05,
+      "loss": 3.5164,
+      "step": 5915
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.28955140767363e-05,
+      "loss": 3.6844,
+      "step": 5920
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.288360009470776e-05,
+      "loss": 3.5802,
+      "step": 5925
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.287167689669683e-05,
+      "loss": 3.5782,
+      "step": 5930
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.285974448526585e-05,
+      "loss": 3.5837,
+      "step": 5935
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.284780286297918e-05,
+      "loss": 3.6449,
+      "step": 5940
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.283585203240317e-05,
+      "loss": 3.6488,
+      "step": 5945
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.282389199610613e-05,
+      "loss": 3.6682,
+      "step": 5950
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.281192275665836e-05,
+      "loss": 3.5109,
+      "step": 5955
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.279994431663212e-05,
+      "loss": 3.5957,
+      "step": 5960
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.278795667860167e-05,
+      "loss": 3.5741,
+      "step": 5965
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.277595984514325e-05,
+      "loss": 3.5312,
+      "step": 5970
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.276395381883502e-05,
+      "loss": 3.5387,
+      "step": 5975
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.275193860225722e-05,
+      "loss": 3.6399,
+      "step": 5980
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.273991419799193e-05,
+      "loss": 3.558,
+      "step": 5985
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.272788060862334e-05,
+      "loss": 3.5936,
+      "step": 5990
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.271583783673751e-05,
+      "loss": 3.5461,
+      "step": 5995
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.270378588492255e-05,
+      "loss": 3.5053,
+      "step": 6000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.269172475576849e-05,
+      "loss": 3.7056,
+      "step": 6005
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.267965445186733e-05,
+      "loss": 3.5454,
+      "step": 6010
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.266757497581308e-05,
+      "loss": 3.4969,
+      "step": 6015
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.265548633020174e-05,
+      "loss": 3.5475,
+      "step": 6020
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.264338851763117e-05,
+      "loss": 3.5754,
+      "step": 6025
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.263128154070133e-05,
+      "loss": 3.7042,
+      "step": 6030
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.261916540201407e-05,
+      "loss": 3.6517,
+      "step": 6035
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.260704010417325e-05,
+      "loss": 3.5246,
+      "step": 6040
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.259490564978466e-05,
+      "loss": 3.4864,
+      "step": 6045
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.258276204145609e-05,
+      "loss": 3.6037,
+      "step": 6050
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.257060928179726e-05,
+      "loss": 3.6137,
+      "step": 6055
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.255844737341992e-05,
+      "loss": 3.5717,
+      "step": 6060
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.254627631893773e-05,
+      "loss": 3.5977,
+      "step": 6065
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.253409612096635e-05,
+      "loss": 3.6441,
+      "step": 6070
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.252190678212335e-05,
+      "loss": 3.5597,
+      "step": 6075
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.250970830502834e-05,
+      "loss": 3.521,
+      "step": 6080
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.249750069230285e-05,
+      "loss": 3.4414,
+      "step": 6085
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.248528394657036e-05,
+      "loss": 3.5877,
+      "step": 6090
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.247305807045637e-05,
+      "loss": 3.5841,
+      "step": 6095
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.246082306658827e-05,
+      "loss": 3.5628,
+      "step": 6100
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.244857893759545e-05,
+      "loss": 3.6324,
+      "step": 6105
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.243632568610929e-05,
+      "loss": 3.5898,
+      "step": 6110
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.242406331476307e-05,
+      "loss": 3.6123,
+      "step": 6115
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.241179182619206e-05,
+      "loss": 3.5837,
+      "step": 6120
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.239951122303349e-05,
+      "loss": 3.6021,
+      "step": 6125
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.238722150792658e-05,
+      "loss": 3.5697,
+      "step": 6130
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.237492268351242e-05,
+      "loss": 3.4668,
+      "step": 6135
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.236261475243415e-05,
+      "loss": 3.6422,
+      "step": 6140
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.23502977173368e-05,
+      "loss": 3.581,
+      "step": 6145
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.233797158086742e-05,
+      "loss": 3.5596,
+      "step": 6150
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.232563634567495e-05,
+      "loss": 3.6226,
+      "step": 6155
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.231329201441034e-05,
+      "loss": 3.5999,
+      "step": 6160
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.230093858972646e-05,
+      "loss": 3.6402,
+      "step": 6165
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.228857607427816e-05,
+      "loss": 3.6862,
+      "step": 6170
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.227620447072221e-05,
+      "loss": 3.672,
+      "step": 6175
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.226382378171738e-05,
+      "loss": 3.6098,
+      "step": 6180
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.225143400992435e-05,
+      "loss": 3.5706,
+      "step": 6185
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.223903515800576e-05,
+      "loss": 3.5565,
+      "step": 6190
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.222662722862622e-05,
+      "loss": 3.6007,
+      "step": 6195
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.221421022445228e-05,
+      "loss": 3.5288,
+      "step": 6200
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.220178414815242e-05,
+      "loss": 3.567,
+      "step": 6205
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.218934900239713e-05,
+      "loss": 3.531,
+      "step": 6210
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.217690478985878e-05,
+      "loss": 3.5394,
+      "step": 6215
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.216445151321171e-05,
+      "loss": 3.6039,
+      "step": 6220
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.215198917513223e-05,
+      "loss": 3.5976,
+      "step": 6225
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.213951777829861e-05,
+      "loss": 3.6074,
+      "step": 6230
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.212703732539098e-05,
+      "loss": 3.487,
+      "step": 6235
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.211454781909153e-05,
+      "loss": 3.5575,
+      "step": 6240
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.21020492620843e-05,
+      "loss": 3.5348,
+      "step": 6245
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.208954165705534e-05,
+      "loss": 3.4605,
+      "step": 6250
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.207702500669262e-05,
+      "loss": 3.4656,
+      "step": 6255
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.206449931368604e-05,
+      "loss": 3.5195,
+      "step": 6260
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.205196458072747e-05,
+      "loss": 3.6519,
+      "step": 6265
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.203942081051071e-05,
+      "loss": 3.506,
+      "step": 6270
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.20268680057315e-05,
+      "loss": 3.5741,
+      "step": 6275
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.201430616908752e-05,
+      "loss": 3.6502,
+      "step": 6280
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.200173530327842e-05,
+      "loss": 3.5597,
+      "step": 6285
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.198915541100575e-05,
+      "loss": 3.5846,
+      "step": 6290
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.1976566494973e-05,
+      "loss": 3.5697,
+      "step": 6295
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.196396855788563e-05,
+      "loss": 3.5557,
+      "step": 6300
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.195136160245104e-05,
+      "loss": 3.5238,
+      "step": 6305
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.193874563137853e-05,
+      "loss": 3.5075,
+      "step": 6310
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.192612064737934e-05,
+      "loss": 3.5903,
+      "step": 6315
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.19134866531667e-05,
+      "loss": 3.5325,
+      "step": 6320
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.190084365145575e-05,
+      "loss": 3.5081,
+      "step": 6325
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.188819164496354e-05,
+      "loss": 3.4329,
+      "step": 6330
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.187553063640906e-05,
+      "loss": 3.5054,
+      "step": 6335
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.186286062851327e-05,
+      "loss": 3.6208,
+      "step": 6340
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.185018162399905e-05,
+      "loss": 3.5159,
+      "step": 6345
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.183749362559118e-05,
+      "loss": 3.5452,
+      "step": 6350
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.182479663601642e-05,
+      "loss": 3.6731,
+      "step": 6355
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.181209065800342e-05,
+      "loss": 3.5044,
+      "step": 6360
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.17993756942828e-05,
+      "loss": 3.5762,
+      "step": 6365
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.17866517475871e-05,
+      "loss": 3.5205,
+      "step": 6370
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.177391882065077e-05,
+      "loss": 3.6241,
+      "step": 6375
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.176117691621017e-05,
+      "loss": 3.5834,
+      "step": 6380
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.17484260370037e-05,
+      "loss": 3.5483,
+      "step": 6385
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.173566618577156e-05,
+      "loss": 3.5598,
+      "step": 6390
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.172289736525594e-05,
+      "loss": 3.5956,
+      "step": 6395
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.171011957820097e-05,
+      "loss": 3.5216,
+      "step": 6400
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.169733282735265e-05,
+      "loss": 3.5531,
+      "step": 6405
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.168453711545895e-05,
+      "loss": 3.6159,
+      "step": 6410
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.167173244526977e-05,
+      "loss": 3.5151,
+      "step": 6415
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.165891881953692e-05,
+      "loss": 3.5704,
+      "step": 6420
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.164609624101414e-05,
+      "loss": 3.5029,
+      "step": 6425
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.163326471245707e-05,
+      "loss": 3.571,
+      "step": 6430
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.162042423662331e-05,
+      "loss": 3.5577,
+      "step": 6435
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.160757481627237e-05,
+      "loss": 3.5597,
+      "step": 6440
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.159471645416567e-05,
+      "loss": 3.6354,
+      "step": 6445
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.158184915306655e-05,
+      "loss": 3.6589,
+      "step": 6450
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.156897291574034e-05,
+      "loss": 3.503,
+      "step": 6455
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.155608774495415e-05,
+      "loss": 3.6651,
+      "step": 6460
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.154319364347715e-05,
+      "loss": 3.5848,
+      "step": 6465
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.153029061408037e-05,
+      "loss": 3.6259,
+      "step": 6470
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.151737865953671e-05,
+      "loss": 3.622,
+      "step": 6475
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.150445778262109e-05,
+      "loss": 3.5562,
+      "step": 6480
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.149152798611027e-05,
+      "loss": 3.5022,
+      "step": 6485
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.147858927278296e-05,
+      "loss": 3.6088,
+      "step": 6490
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.14656416454198e-05,
+      "loss": 3.639,
+      "step": 6495
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.14526851068033e-05,
+      "loss": 3.5439,
+      "step": 6500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.143971965971789e-05,
+      "loss": 3.534,
+      "step": 6505
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.142674530694999e-05,
+      "loss": 3.4733,
+      "step": 6510
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.141376205128784e-05,
+      "loss": 3.7137,
+      "step": 6515
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.140076989552162e-05,
+      "loss": 3.5804,
+      "step": 6520
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.138776884244347e-05,
+      "loss": 3.4639,
+      "step": 6525
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.137475889484739e-05,
+      "loss": 3.6344,
+      "step": 6530
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.13617400555293e-05,
+      "loss": 3.4193,
+      "step": 6535
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.134871232728706e-05,
+      "loss": 3.537,
+      "step": 6540
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.133567571292038e-05,
+      "loss": 3.5168,
+      "step": 6545
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.132263021523096e-05,
+      "loss": 3.4677,
+      "step": 6550
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.130957583702236e-05,
+      "loss": 3.5843,
+      "step": 6555
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.129651258110003e-05,
+      "loss": 3.5941,
+      "step": 6560
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.128344045027137e-05,
+      "loss": 3.5584,
+      "step": 6565
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.127035944734568e-05,
+      "loss": 3.5377,
+      "step": 6570
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.125726957513414e-05,
+      "loss": 3.4998,
+      "step": 6575
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.124417083644988e-05,
+      "loss": 3.544,
+      "step": 6580
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.12310632341079e-05,
+      "loss": 3.5726,
+      "step": 6585
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.121794677092509e-05,
+      "loss": 3.6388,
+      "step": 6590
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.120482144972029e-05,
+      "loss": 3.5607,
+      "step": 6595
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.119168727331424e-05,
+      "loss": 3.4596,
+      "step": 6600
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.117854424452952e-05,
+      "loss": 3.5196,
+      "step": 6605
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.116539236619071e-05,
+      "loss": 3.5832,
+      "step": 6610
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.115223164112418e-05,
+      "loss": 3.4216,
+      "step": 6615
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.113906207215832e-05,
+      "loss": 3.4464,
+      "step": 6620
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.112588366212333e-05,
+      "loss": 3.5169,
+      "step": 6625
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.111269641385134e-05,
+      "loss": 3.6711,
+      "step": 6630
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.10995003301764e-05,
+      "loss": 3.4761,
+      "step": 6635
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.10862954139344e-05,
+      "loss": 3.5652,
+      "step": 6640
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.107308166796322e-05,
+      "loss": 3.6506,
+      "step": 6645
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.105985909510255e-05,
+      "loss": 3.6716,
+      "step": 6650
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.104662769819404e-05,
+      "loss": 3.6558,
+      "step": 6655
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.103338748008118e-05,
+      "loss": 3.6856,
+      "step": 6660
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.102013844360939e-05,
+      "loss": 3.5786,
+      "step": 6665
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.100688059162599e-05,
+      "loss": 3.4324,
+      "step": 6670
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.099361392698018e-05,
+      "loss": 3.6528,
+      "step": 6675
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.098033845252307e-05,
+      "loss": 3.5293,
+      "step": 6680
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.096705417110762e-05,
+      "loss": 3.5525,
+      "step": 6685
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.095376108558875e-05,
+      "loss": 3.4214,
+      "step": 6690
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.094045919882322e-05,
+      "loss": 3.6658,
+      "step": 6695
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.092714851366968e-05,
+      "loss": 3.606,
+      "step": 6700
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.091382903298872e-05,
+      "loss": 3.5723,
+      "step": 6705
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.090050075964276e-05,
+      "loss": 3.5262,
+      "step": 6710
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.088716369649616e-05,
+      "loss": 3.4272,
+      "step": 6715
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.087381784641515e-05,
+      "loss": 3.5084,
+      "step": 6720
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.086046321226783e-05,
+      "loss": 3.5122,
+      "step": 6725
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.08470997969242e-05,
+      "loss": 3.4654,
+      "step": 6730
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.083372760325617e-05,
+      "loss": 3.5861,
+      "step": 6735
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.08203466341375e-05,
+      "loss": 3.551,
+      "step": 6740
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.080695689244385e-05,
+      "loss": 3.5268,
+      "step": 6745
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.079355838105279e-05,
+      "loss": 3.4525,
+      "step": 6750
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.078015110284373e-05,
+      "loss": 3.3945,
+      "step": 6755
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.076673506069801e-05,
+      "loss": 3.514,
+      "step": 6760
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.07533102574988e-05,
+      "loss": 3.5778,
+      "step": 6765
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.07398766961312e-05,
+      "loss": 3.404,
+      "step": 6770
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.072643437948218e-05,
+      "loss": 3.7109,
+      "step": 6775
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.07129833104406e-05,
+      "loss": 3.5104,
+      "step": 6780
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.069952349189714e-05,
+      "loss": 3.4361,
+      "step": 6785
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.068605492674446e-05,
+      "loss": 3.4596,
+      "step": 6790
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.0672577617877e-05,
+      "loss": 3.4031,
+      "step": 6795
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.065909156819118e-05,
+      "loss": 3.572,
+      "step": 6800
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.06455967805852e-05,
+      "loss": 3.5457,
+      "step": 6805
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.063209325795921e-05,
+      "loss": 3.5093,
+      "step": 6810
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.06185810032152e-05,
+      "loss": 3.5327,
+      "step": 6815
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.060506001925707e-05,
+      "loss": 3.4918,
+      "step": 6820
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.059153030899051e-05,
+      "loss": 3.6342,
+      "step": 6825
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.05779918753232e-05,
+      "loss": 3.5013,
+      "step": 6830
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.056444472116465e-05,
+      "loss": 3.5591,
+      "step": 6835
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.055088884942622e-05,
+      "loss": 3.5687,
+      "step": 6840
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.053732426302116e-05,
+      "loss": 3.5036,
+      "step": 6845
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.052375096486458e-05,
+      "loss": 3.4463,
+      "step": 6850
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.051016895787352e-05,
+      "loss": 3.5371,
+      "step": 6855
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.049657824496678e-05,
+      "loss": 3.3967,
+      "step": 6860
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.048297882906516e-05,
+      "loss": 3.5272,
+      "step": 6865
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.046937071309124e-05,
+      "loss": 3.5581,
+      "step": 6870
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.04557538999695e-05,
+      "loss": 3.5465,
+      "step": 6875
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.04421283926263e-05,
+      "loss": 3.6113,
+      "step": 6880
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.042849419398982e-05,
+      "loss": 3.4789,
+      "step": 6885
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.041485130699018e-05,
+      "loss": 3.6647,
+      "step": 6890
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.040119973455934e-05,
+      "loss": 3.5378,
+      "step": 6895
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.038753947963108e-05,
+      "loss": 3.444,
+      "step": 6900
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.03738705451411e-05,
+      "loss": 3.5048,
+      "step": 6905
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.036019293402697e-05,
+      "loss": 3.5103,
+      "step": 6910
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.034650664922807e-05,
+      "loss": 3.6116,
+      "step": 6915
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.033281169368568e-05,
+      "loss": 3.5288,
+      "step": 6920
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.031910807034296e-05,
+      "loss": 3.4385,
+      "step": 6925
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.03053957821449e-05,
+      "loss": 3.586,
+      "step": 6930
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.029167483203837e-05,
+      "loss": 3.5335,
+      "step": 6935
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.027794522297211e-05,
+      "loss": 3.5654,
+      "step": 6940
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.026420695789668e-05,
+      "loss": 3.4089,
+      "step": 6945
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.025046003976457e-05,
+      "loss": 3.54,
+      "step": 6950
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.023670447153003e-05,
+      "loss": 3.5416,
+      "step": 6955
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.022294025614926e-05,
+      "loss": 3.4699,
+      "step": 6960
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.02091673965803e-05,
+      "loss": 3.5675,
+      "step": 6965
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.0195385895783e-05,
+      "loss": 3.5741,
+      "step": 6970
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.018159575671914e-05,
+      "loss": 3.5114,
+      "step": 6975
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.016779698235227e-05,
+      "loss": 3.5919,
+      "step": 6980
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.015398957564787e-05,
+      "loss": 3.5126,
+      "step": 6985
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.014017353957324e-05,
+      "loss": 3.5685,
+      "step": 6990
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.012634887709754e-05,
+      "loss": 3.532,
+      "step": 6995
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.01125155911918e-05,
+      "loss": 3.5481,
+      "step": 7000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.009867368482887e-05,
+      "loss": 3.603,
+      "step": 7005
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.008482316098348e-05,
+      "loss": 3.5523,
+      "step": 7010
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.007096402263221e-05,
+      "loss": 3.3839,
+      "step": 7015
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.005709627275348e-05,
+      "loss": 3.4187,
+      "step": 7020
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.004321991432753e-05,
+      "loss": 3.5402,
+      "step": 7025
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.002933495033657e-05,
+      "loss": 3.4514,
+      "step": 7030
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.001544138376447e-05,
+      "loss": 3.5998,
+      "step": 7035
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.000153921759714e-05,
+      "loss": 3.5431,
+      "step": 7040
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.998762845482223e-05,
+      "loss": 3.4862,
+      "step": 7045
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.997370909842921e-05,
+      "loss": 3.4169,
+      "step": 7050
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.99597811514095e-05,
+      "loss": 3.4929,
+      "step": 7055
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.99458446167563e-05,
+      "loss": 3.533,
+      "step": 7060
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.993189949746468e-05,
+      "loss": 3.5534,
+      "step": 7065
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.99179457965315e-05,
+      "loss": 3.5099,
+      "step": 7070
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.990398351695555e-05,
+      "loss": 3.5415,
+      "step": 7075
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.989001266173738e-05,
+      "loss": 3.5478,
+      "step": 7080
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.987603323387948e-05,
+      "loss": 3.5151,
+      "step": 7085
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.986204523638606e-05,
+      "loss": 3.4512,
+      "step": 7090
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.984804867226328e-05,
+      "loss": 3.4515,
+      "step": 7095
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.98340435445191e-05,
+      "loss": 3.4756,
+      "step": 7100
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.982002985616331e-05,
+      "loss": 3.6139,
+      "step": 7105
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.980600761020754e-05,
+      "loss": 3.419,
+      "step": 7110
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.979197680966527e-05,
+      "loss": 3.5376,
+      "step": 7115
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.977793745755182e-05,
+      "loss": 3.5058,
+      "step": 7120
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.976388955688434e-05,
+      "loss": 3.529,
+      "step": 7125
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.974983311068184e-05,
+      "loss": 3.4919,
+      "step": 7130
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.973576812196513e-05,
+      "loss": 3.4312,
+      "step": 7135
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.972169459375686e-05,
+      "loss": 3.428,
+      "step": 7140
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.970761252908156e-05,
+      "loss": 3.541,
+      "step": 7145
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.969352193096554e-05,
+      "loss": 3.4676,
+      "step": 7150
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.967942280243699e-05,
+      "loss": 3.4745,
+      "step": 7155
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.966531514652588e-05,
+      "loss": 3.4614,
+      "step": 7160
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.965119896626406e-05,
+      "loss": 3.4995,
+      "step": 7165
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.963707426468519e-05,
+      "loss": 3.5288,
+      "step": 7170
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.962294104482478e-05,
+      "loss": 3.4912,
+      "step": 7175
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.960879930972014e-05,
+      "loss": 3.4556,
+      "step": 7180
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.959464906241044e-05,
+      "loss": 3.5424,
+      "step": 7185
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.958049030593665e-05,
+      "loss": 3.441,
+      "step": 7190
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.956632304334158e-05,
+      "loss": 3.5728,
+      "step": 7195
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.955214727766989e-05,
+      "loss": 3.5114,
+      "step": 7200
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.953796301196807e-05,
+      "loss": 3.4704,
+      "step": 7205
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.952377024928437e-05,
+      "loss": 3.6184,
+      "step": 7210
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.950956899266895e-05,
+      "loss": 3.5997,
+      "step": 7215
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.949535924517374e-05,
+      "loss": 3.4919,
+      "step": 7220
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.948114100985252e-05,
+      "loss": 3.5112,
+      "step": 7225
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.946691428976088e-05,
+      "loss": 3.4752,
+      "step": 7230
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.945267908795625e-05,
+      "loss": 3.4226,
+      "step": 7235
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.943843540749785e-05,
+      "loss": 3.4503,
+      "step": 7240
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.942418325144679e-05,
+      "loss": 3.5223,
+      "step": 7245
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.940992262286592e-05,
+      "loss": 3.5969,
+      "step": 7250
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.939565352481996e-05,
+      "loss": 3.5297,
+      "step": 7255
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.938137596037544e-05,
+      "loss": 3.5613,
+      "step": 7260
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.93670899326007e-05,
+      "loss": 3.4675,
+      "step": 7265
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.93527954445659e-05,
+      "loss": 3.5134,
+      "step": 7270
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.933849249934306e-05,
+      "loss": 3.486,
+      "step": 7275
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.932418110000594e-05,
+      "loss": 3.58,
+      "step": 7280
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.930986124963016e-05,
+      "loss": 3.4639,
+      "step": 7285
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.92955329512932e-05,
+      "loss": 3.4393,
+      "step": 7290
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.928119620807426e-05,
+      "loss": 3.4737,
+      "step": 7295
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.926685102305443e-05,
+      "loss": 3.559,
+      "step": 7300
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.925249739931659e-05,
+      "loss": 3.5177,
+      "step": 7305
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.923813533994543e-05,
+      "loss": 3.4876,
+      "step": 7310
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.922376484802745e-05,
+      "loss": 3.5773,
+      "step": 7315
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.920938592665098e-05,
+      "loss": 3.434,
+      "step": 7320
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.919499857890613e-05,
+      "loss": 3.5259,
+      "step": 7325
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.918060280788486e-05,
+      "loss": 3.498,
+      "step": 7330
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.916619861668092e-05,
+      "loss": 3.4358,
+      "step": 7335
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.915178600838985e-05,
+      "loss": 3.5804,
+      "step": 7340
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.913736498610906e-05,
+      "loss": 3.5247,
+      "step": 7345
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.912293555293767e-05,
+      "loss": 3.4963,
+      "step": 7350
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.910849771197671e-05,
+      "loss": 3.5073,
+      "step": 7355
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.909405146632896e-05,
+      "loss": 3.4326,
+      "step": 7360
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.907959681909902e-05,
+      "loss": 3.4629,
+      "step": 7365
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.906513377339327e-05,
+      "loss": 3.4847,
+      "step": 7370
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.905066233231995e-05,
+      "loss": 3.5267,
+      "step": 7375
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.903618249898906e-05,
+      "loss": 3.5427,
+      "step": 7380
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.902169427651242e-05,
+      "loss": 3.4486,
+      "step": 7385
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.900719766800365e-05,
+      "loss": 3.5533,
+      "step": 7390
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.899269267657817e-05,
+      "loss": 3.4521,
+      "step": 7395
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.89781793053532e-05,
+      "loss": 3.3999,
+      "step": 7400
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.896365755744775e-05,
+      "loss": 3.4211,
+      "step": 7405
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.894912743598268e-05,
+      "loss": 3.3688,
+      "step": 7410
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.893458894408058e-05,
+      "loss": 3.4706,
+      "step": 7415
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.89200420848659e-05,
+      "loss": 3.4638,
+      "step": 7420
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.890548686146483e-05,
+      "loss": 3.4547,
+      "step": 7425
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.889092327700541e-05,
+      "loss": 3.611,
+      "step": 7430
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.887635133461745e-05,
+      "loss": 3.5858,
+      "step": 7435
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.886177103743257e-05,
+      "loss": 3.4523,
+      "step": 7440
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.884718238858416e-05,
+      "loss": 3.3612,
+      "step": 7445
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.883258539120742e-05,
+      "loss": 3.4537,
+      "step": 7450
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.881798004843938e-05,
+      "loss": 3.397,
+      "step": 7455
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.880336636341878e-05,
+      "loss": 3.3828,
+      "step": 7460
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.878874433928623e-05,
+      "loss": 3.4596,
+      "step": 7465
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.877411397918411e-05,
+      "loss": 3.5192,
+      "step": 7470
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.875947528625658e-05,
+      "loss": 3.5118,
+      "step": 7475
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.874482826364958e-05,
+      "loss": 3.5133,
+      "step": 7480
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.873017291451086e-05,
+      "loss": 3.4795,
+      "step": 7485
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.871550924198999e-05,
+      "loss": 3.5096,
+      "step": 7490
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.870083724923826e-05,
+      "loss": 3.4943,
+      "step": 7495
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.868615693940879e-05,
+      "loss": 3.6153,
+      "step": 7500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.86714683156565e-05,
+      "loss": 3.5453,
+      "step": 7505
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.865677138113803e-05,
+      "loss": 3.514,
+      "step": 7510
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.864206613901189e-05,
+      "loss": 3.509,
+      "step": 7515
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.862735259243832e-05,
+      "loss": 3.4351,
+      "step": 7520
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.861263074457941e-05,
+      "loss": 3.573,
+      "step": 7525
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.859790059859893e-05,
+      "loss": 3.4291,
+      "step": 7530
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.858316215766252e-05,
+      "loss": 3.5962,
+      "step": 7535
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.856841542493754e-05,
+      "loss": 3.6117,
+      "step": 7540
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.855366040359321e-05,
+      "loss": 3.3975,
+      "step": 7545
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.853889709680045e-05,
+      "loss": 3.4811,
+      "step": 7550
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.852412550773202e-05,
+      "loss": 3.542,
+      "step": 7555
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.850934563956244e-05,
+      "loss": 3.5076,
+      "step": 7560
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.849455749546797e-05,
+      "loss": 3.4975,
+      "step": 7565
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.847976107862673e-05,
+      "loss": 3.5067,
+      "step": 7570
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.846495639221854e-05,
+      "loss": 3.5526,
+      "step": 7575
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.845014343942505e-05,
+      "loss": 3.5319,
+      "step": 7580
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.843532222342965e-05,
+      "loss": 3.3806,
+      "step": 7585
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.842049274741753e-05,
+      "loss": 3.5225,
+      "step": 7590
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.840565501457565e-05,
+      "loss": 3.5024,
+      "step": 7595
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.839080902809274e-05,
+      "loss": 3.4989,
+      "step": 7600
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.83759547911593e-05,
+      "loss": 3.5711,
+      "step": 7605
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.83610923069676e-05,
+      "loss": 3.4634,
+      "step": 7610
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.83462215787117e-05,
+      "loss": 3.4583,
+      "step": 7615
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.833134260958745e-05,
+      "loss": 3.5227,
+      "step": 7620
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.831645540279238e-05,
+      "loss": 3.4591,
+      "step": 7625
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.830155996152591e-05,
+      "loss": 3.4781,
+      "step": 7630
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.828665628898918e-05,
+      "loss": 3.4303,
+      "step": 7635
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.827174438838503e-05,
+      "loss": 3.4305,
+      "step": 7640
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.825682426291819e-05,
+      "loss": 3.4548,
+      "step": 7645
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.824189591579505e-05,
+      "loss": 3.3811,
+      "step": 7650
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.822695935022387e-05,
+      "loss": 3.5081,
+      "step": 7655
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.821201456941458e-05,
+      "loss": 3.5186,
+      "step": 7660
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.819706157657893e-05,
+      "loss": 3.4659,
+      "step": 7665
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.818210037493042e-05,
+      "loss": 3.3918,
+      "step": 7670
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.816713096768431e-05,
+      "loss": 3.3487,
+      "step": 7675
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.815215335805765e-05,
+      "loss": 3.5102,
+      "step": 7680
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.81371675492692e-05,
+      "loss": 3.4507,
+      "step": 7685
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.812217354453956e-05,
+      "loss": 3.4315,
+      "step": 7690
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.810717134709099e-05,
+      "loss": 3.484,
+      "step": 7695
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.809216096014759e-05,
+      "loss": 3.5864,
+      "step": 7700
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.807714238693521e-05,
+      "loss": 3.4511,
+      "step": 7705
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.806211563068145e-05,
+      "loss": 3.5034,
+      "step": 7710
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.804708069461563e-05,
+      "loss": 3.5788,
+      "step": 7715
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.803203758196887e-05,
+      "loss": 3.4799,
+      "step": 7720
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.801698629597407e-05,
+      "loss": 3.5083,
+      "step": 7725
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.800192683986583e-05,
+      "loss": 3.4058,
+      "step": 7730
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.798685921688053e-05,
+      "loss": 3.4338,
+      "step": 7735
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.797178343025631e-05,
+      "loss": 3.4857,
+      "step": 7740
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.795669948323308e-05,
+      "loss": 3.5212,
+      "step": 7745
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.794160737905245e-05,
+      "loss": 3.3967,
+      "step": 7750
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.792650712095784e-05,
+      "loss": 3.3719,
+      "step": 7755
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.791139871219441e-05,
+      "loss": 3.4683,
+      "step": 7760
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.789628215600902e-05,
+      "loss": 3.4988,
+      "step": 7765
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.788115745565036e-05,
+      "loss": 3.5674,
+      "step": 7770
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.786602461436884e-05,
+      "loss": 3.466,
+      "step": 7775
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.785088363541657e-05,
+      "loss": 3.4507,
+      "step": 7780
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.78357345220475e-05,
+      "loss": 3.4228,
+      "step": 7785
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.782057727751722e-05,
+      "loss": 3.4264,
+      "step": 7790
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.78054119050832e-05,
+      "loss": 3.3594,
+      "step": 7795
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.779023840800452e-05,
+      "loss": 3.4017,
+      "step": 7800
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.777505678954211e-05,
+      "loss": 3.3921,
+      "step": 7805
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.775986705295858e-05,
+      "loss": 3.4618,
+      "step": 7810
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.774466920151833e-05,
+      "loss": 3.4593,
+      "step": 7815
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.772946323848746e-05,
+      "loss": 3.2764,
+      "step": 7820
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.771424916713386e-05,
+      "loss": 3.5471,
+      "step": 7825
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.769902699072712e-05,
+      "loss": 3.4716,
+      "step": 7830
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.768379671253859e-05,
+      "loss": 3.4505,
+      "step": 7835
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.76685583358414e-05,
+      "loss": 3.645,
+      "step": 7840
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.765331186391033e-05,
+      "loss": 3.4364,
+      "step": 7845
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.763805730002199e-05,
+      "loss": 3.528,
+      "step": 7850
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.762279464745466e-05,
+      "loss": 3.4277,
+      "step": 7855
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.760752390948843e-05,
+      "loss": 3.4136,
+      "step": 7860
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.759224508940505e-05,
+      "loss": 3.4857,
+      "step": 7865
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.757695819048808e-05,
+      "loss": 3.4008,
+      "step": 7870
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.756166321602273e-05,
+      "loss": 3.498,
+      "step": 7875
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.754636016929604e-05,
+      "loss": 3.4034,
+      "step": 7880
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.753104905359674e-05,
+      "loss": 3.468,
+      "step": 7885
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.751572987221526e-05,
+      "loss": 3.4808,
+      "step": 7890
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.750040262844384e-05,
+      "loss": 3.4407,
+      "step": 7895
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.748506732557637e-05,
+      "loss": 3.3546,
+      "step": 7900
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.746972396690853e-05,
+      "loss": 3.4793,
+      "step": 7905
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.745437255573773e-05,
+      "loss": 3.4071,
+      "step": 7910
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.743901309536307e-05,
+      "loss": 3.5352,
+      "step": 7915
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.742364558908543e-05,
+      "loss": 3.358,
+      "step": 7920
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.740827004020735e-05,
+      "loss": 3.4655,
+      "step": 7925
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.739288645203319e-05,
+      "loss": 3.5018,
+      "step": 7930
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.737749482786897e-05,
+      "loss": 3.5549,
+      "step": 7935
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.736209517102247e-05,
+      "loss": 3.6071,
+      "step": 7940
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.734668748480315e-05,
+      "loss": 3.4605,
+      "step": 7945
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.733127177252225e-05,
+      "loss": 3.4953,
+      "step": 7950
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.731584803749271e-05,
+      "loss": 3.3951,
+      "step": 7955
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.73004162830292e-05,
+      "loss": 3.3067,
+      "step": 7960
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.728497651244812e-05,
+      "loss": 3.3933,
+      "step": 7965
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.726952872906755e-05,
+      "loss": 3.5993,
+      "step": 7970
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.725407293620737e-05,
+      "loss": 3.5275,
+      "step": 7975
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.72386091371891e-05,
+      "loss": 3.4345,
+      "step": 7980
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.722313733533603e-05,
+      "loss": 3.4515,
+      "step": 7985
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.720765753397317e-05,
+      "loss": 3.4055,
+      "step": 7990
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.719216973642723e-05,
+      "loss": 3.3859,
+      "step": 7995
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.717667394602664e-05,
+      "loss": 3.5479,
+      "step": 8000
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 33984,
+  "num_train_epochs": 1,
+  "save_steps": 1000,
+  "total_flos": 2.208473472093389e+19,
+  "trial_name": null,
+  "trial_params": null
+}