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": 0.20764523811068789,
+  "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": 18617.1266,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1e-05,
+      "loss": 12423.2211,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5e-05,
+      "loss": 5428.3555,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2e-05,
+      "loss": 3044.3764,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5e-05,
+      "loss": 1740.9174,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3e-05,
+      "loss": 892.687,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.5e-05,
+      "loss": 442.4878,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4e-05,
+      "loss": 229.281,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.5e-05,
+      "loss": 103.0141,
+      "step": 45
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5e-05,
+      "loss": 49.7704,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.500000000000001e-05,
+      "loss": 32.9897,
+      "step": 55
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6e-05,
+      "loss": 15.7778,
+      "step": 60
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.500000000000001e-05,
+      "loss": 9.5009,
+      "step": 65
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7e-05,
+      "loss": 8.4248,
+      "step": 70
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 8.3156,
+      "step": 75
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8e-05,
+      "loss": 8.1767,
+      "step": 80
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.5e-05,
+      "loss": 8.098,
+      "step": 85
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9e-05,
+      "loss": 8.1834,
+      "step": 90
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.5e-05,
+      "loss": 9.9323,
+      "step": 95
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001,
+      "loss": 8.201,
+      "step": 100
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999999582259295e-05,
+      "loss": 8.2426,
+      "step": 105
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999998329037248e-05,
+      "loss": 8.1147,
+      "step": 110
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999996240334068e-05,
+      "loss": 8.0152,
+      "step": 115
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999993316150106e-05,
+      "loss": 7.7808,
+      "step": 120
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999989556485852e-05,
+      "loss": 7.771,
+      "step": 125
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999984961341928e-05,
+      "loss": 7.7533,
+      "step": 130
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999979530719108e-05,
+      "loss": 7.6443,
+      "step": 135
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999973264618297e-05,
+      "loss": 7.5425,
+      "step": 140
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999966163040541e-05,
+      "loss": 7.6104,
+      "step": 145
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.99995822598703e-05,
+      "loss": 7.522,
+      "step": 150
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999949453459088e-05,
+      "loss": 7.5905,
+      "step": 155
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999939845458179e-05,
+      "loss": 7.3457,
+      "step": 160
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999929401985911e-05,
+      "loss": 7.3924,
+      "step": 165
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.99991812304403e-05,
+      "loss": 7.2623,
+      "step": 170
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999906008634416e-05,
+      "loss": 7.2625,
+      "step": 175
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.9998930587591e-05,
+      "loss": 7.2124,
+      "step": 180
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.99987927342024e-05,
+      "loss": 7.1701,
+      "step": 185
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999864652620143e-05,
+      "loss": 7.0713,
+      "step": 190
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999849196361251e-05,
+      "loss": 6.9876,
+      "step": 195
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999832904646147e-05,
+      "loss": 6.921,
+      "step": 200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99981577747755e-05,
+      "loss": 6.9449,
+      "step": 205
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999797814858327e-05,
+      "loss": 6.8032,
+      "step": 210
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999779016791477e-05,
+      "loss": 6.6663,
+      "step": 215
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999759383280141e-05,
+      "loss": 6.6613,
+      "step": 220
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.9997389143276e-05,
+      "loss": 6.6939,
+      "step": 225
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999717609937275e-05,
+      "loss": 6.5651,
+      "step": 230
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999695470112725e-05,
+      "loss": 6.572,
+      "step": 235
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99967249485765e-05,
+      "loss": 6.5009,
+      "step": 240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999648684175888e-05,
+      "loss": 6.5609,
+      "step": 245
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999624038071418e-05,
+      "loss": 6.49,
+      "step": 250
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99959855654836e-05,
+      "loss": 6.4226,
+      "step": 255
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999572239610968e-05,
+      "loss": 6.3265,
+      "step": 260
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999545087263645e-05,
+      "loss": 6.2756,
+      "step": 265
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999517099510923e-05,
+      "loss": 6.3017,
+      "step": 270
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999488276357481e-05,
+      "loss": 6.4622,
+      "step": 275
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999458617808137e-05,
+      "loss": 6.2479,
+      "step": 280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999428123867843e-05,
+      "loss": 6.2877,
+      "step": 285
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999396794541695e-05,
+      "loss": 6.3512,
+      "step": 290
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999364629834931e-05,
+      "loss": 6.1547,
+      "step": 295
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999331629752924e-05,
+      "loss": 6.1545,
+      "step": 300
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999297794301188e-05,
+      "loss": 6.0496,
+      "step": 305
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999263123485376e-05,
+      "loss": 6.0169,
+      "step": 310
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999227617311282e-05,
+      "loss": 6.079,
+      "step": 315
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99919127578484e-05,
+      "loss": 6.1394,
+      "step": 320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999154098912122e-05,
+      "loss": 5.9373,
+      "step": 325
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999116086699338e-05,
+      "loss": 5.9792,
+      "step": 330
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999077239152843e-05,
+      "loss": 6.1228,
+      "step": 335
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999037556279127e-05,
+      "loss": 5.8702,
+      "step": 340
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998997038084818e-05,
+      "loss": 5.8515,
+      "step": 345
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998955684576692e-05,
+      "loss": 5.8557,
+      "step": 350
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998913495761654e-05,
+      "loss": 5.8982,
+      "step": 355
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998870471646757e-05,
+      "loss": 5.7548,
+      "step": 360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99882661223919e-05,
+      "loss": 5.9184,
+      "step": 365
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998781917546278e-05,
+      "loss": 5.7754,
+      "step": 370
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998736387575493e-05,
+      "loss": 5.661,
+      "step": 375
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998690022334442e-05,
+      "loss": 5.7434,
+      "step": 380
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998642821830873e-05,
+      "loss": 5.7023,
+      "step": 385
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99859478607267e-05,
+      "loss": 5.8877,
+      "step": 390
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998545915067864e-05,
+      "loss": 5.7972,
+      "step": 395
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998496208824618e-05,
+      "loss": 5.6729,
+      "step": 400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99844566735124e-05,
+      "loss": 5.6854,
+      "step": 405
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998394290656172e-05,
+      "loss": 5.7168,
+      "step": 410
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998342078748002e-05,
+      "loss": 5.5127,
+      "step": 415
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998289031635453e-05,
+      "loss": 5.541,
+      "step": 420
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998235149327389e-05,
+      "loss": 5.6185,
+      "step": 425
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998180431832816e-05,
+      "loss": 5.6035,
+      "step": 430
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998124879160872e-05,
+      "loss": 5.5598,
+      "step": 435
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998068491320844e-05,
+      "loss": 5.5578,
+      "step": 440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998011268322152e-05,
+      "loss": 5.5256,
+      "step": 445
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997953210174361e-05,
+      "loss": 5.5566,
+      "step": 450
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997894316887167e-05,
+      "loss": 5.5645,
+      "step": 455
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997834588470414e-05,
+      "loss": 5.4581,
+      "step": 460
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997774024934082e-05,
+      "loss": 5.5094,
+      "step": 465
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997712626288292e-05,
+      "loss": 5.5149,
+      "step": 470
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997650392543303e-05,
+      "loss": 5.3256,
+      "step": 475
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997587323709512e-05,
+      "loss": 5.404,
+      "step": 480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997523419797459e-05,
+      "loss": 5.5746,
+      "step": 485
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997458680817824e-05,
+      "loss": 5.3537,
+      "step": 490
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997393106781421e-05,
+      "loss": 5.4468,
+      "step": 495
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99732669769921e-05,
+      "loss": 5.3541,
+      "step": 500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997259453582286e-05,
+      "loss": 5.4529,
+      "step": 505
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997191374441887e-05,
+      "loss": 5.3722,
+      "step": 510
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997122460289387e-05,
+      "loss": 5.3657,
+      "step": 515
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997052711136301e-05,
+      "loss": 5.3573,
+      "step": 520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996982126994287e-05,
+      "loss": 5.3667,
+      "step": 525
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996910707875136e-05,
+      "loss": 5.5042,
+      "step": 530
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996838453790784e-05,
+      "loss": 5.6713,
+      "step": 535
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996765364753302e-05,
+      "loss": 5.3809,
+      "step": 540
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996691440774906e-05,
+      "loss": 5.3201,
+      "step": 545
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996616681867945e-05,
+      "loss": 5.3227,
+      "step": 550
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996541088044913e-05,
+      "loss": 5.3299,
+      "step": 555
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996464659318442e-05,
+      "loss": 5.3159,
+      "step": 560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996387395701302e-05,
+      "loss": 5.4261,
+      "step": 565
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996309297206403e-05,
+      "loss": 5.267,
+      "step": 570
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996230363846797e-05,
+      "loss": 5.4374,
+      "step": 575
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.996150595635671e-05,
+      "loss": 5.4541,
+      "step": 580
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.996069992586355e-05,
+      "loss": 5.3519,
+      "step": 585
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.995988554712318e-05,
+      "loss": 5.3107,
+      "step": 590
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.995906282027166e-05,
+      "loss": 5.3757,
+      "step": 595
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.995823174544649e-05,
+      "loss": 5.4783,
+      "step": 600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.995739232278654e-05,
+      "loss": 5.3228,
+      "step": 605
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.995654455243204e-05,
+      "loss": 5.3228,
+      "step": 610
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.995568843452468e-05,
+      "loss": 5.2705,
+      "step": 615
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99548239692075e-05,
+      "loss": 5.162,
+      "step": 620
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.995395115662498e-05,
+      "loss": 5.1186,
+      "step": 625
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.995306999692293e-05,
+      "loss": 5.1657,
+      "step": 630
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.995218049024858e-05,
+      "loss": 5.099,
+      "step": 635
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.995128263675058e-05,
+      "loss": 4.9959,
+      "step": 640
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.995037643657899e-05,
+      "loss": 5.0853,
+      "step": 645
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994946188988516e-05,
+      "loss": 5.1226,
+      "step": 650
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994853899682197e-05,
+      "loss": 5.5214,
+      "step": 655
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994760775754362e-05,
+      "loss": 5.2146,
+      "step": 660
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994666817220571e-05,
+      "loss": 5.1712,
+      "step": 665
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994572024096521e-05,
+      "loss": 5.1258,
+      "step": 670
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994476396398055e-05,
+      "loss": 5.2082,
+      "step": 675
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994379934141153e-05,
+      "loss": 5.2147,
+      "step": 680
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994282637341932e-05,
+      "loss": 5.0742,
+      "step": 685
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994184506016651e-05,
+      "loss": 5.4435,
+      "step": 690
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994085540181703e-05,
+      "loss": 5.1636,
+      "step": 695
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99398573985363e-05,
+      "loss": 5.1475,
+      "step": 700
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993885105049107e-05,
+      "loss": 5.0508,
+      "step": 705
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993783635784947e-05,
+      "loss": 5.0738,
+      "step": 710
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993681332078108e-05,
+      "loss": 5.1189,
+      "step": 715
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993578193945685e-05,
+      "loss": 5.0099,
+      "step": 720
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99347422140491e-05,
+      "loss": 5.0021,
+      "step": 725
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993369414473157e-05,
+      "loss": 5.0946,
+      "step": 730
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993263773167939e-05,
+      "loss": 5.1012,
+      "step": 735
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993157297506907e-05,
+      "loss": 5.2546,
+      "step": 740
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993049987507856e-05,
+      "loss": 5.0586,
+      "step": 745
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992941843188715e-05,
+      "loss": 5.05,
+      "step": 750
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992832864567554e-05,
+      "loss": 4.9387,
+      "step": 755
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992723051662584e-05,
+      "loss": 5.0998,
+      "step": 760
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992612404492153e-05,
+      "loss": 4.9158,
+      "step": 765
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992500923074752e-05,
+      "loss": 5.059,
+      "step": 770
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992388607429006e-05,
+      "loss": 4.9558,
+      "step": 775
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992275457573685e-05,
+      "loss": 4.9074,
+      "step": 780
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992161473527695e-05,
+      "loss": 4.8895,
+      "step": 785
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992046655310083e-05,
+      "loss": 4.9403,
+      "step": 790
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991931002940034e-05,
+      "loss": 4.9595,
+      "step": 795
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991814516436873e-05,
+      "loss": 4.9005,
+      "step": 800
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991697195820065e-05,
+      "loss": 4.8726,
+      "step": 805
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991579041109212e-05,
+      "loss": 4.9425,
+      "step": 810
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991460052324061e-05,
+      "loss": 4.8039,
+      "step": 815
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991340229484491e-05,
+      "loss": 5.0399,
+      "step": 820
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991219572610526e-05,
+      "loss": 4.9526,
+      "step": 825
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991098081722325e-05,
+      "loss": 4.9462,
+      "step": 830
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990975756840193e-05,
+      "loss": 4.8922,
+      "step": 835
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990852597984566e-05,
+      "loss": 4.9382,
+      "step": 840
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990728605176025e-05,
+      "loss": 4.944,
+      "step": 845
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990603778435288e-05,
+      "loss": 4.8664,
+      "step": 850
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990478117783214e-05,
+      "loss": 4.8651,
+      "step": 855
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990351623240799e-05,
+      "loss": 4.9817,
+      "step": 860
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990224294829181e-05,
+      "loss": 5.0075,
+      "step": 865
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990096132569636e-05,
+      "loss": 4.7549,
+      "step": 870
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.98996713648358e-05,
+      "loss": 4.8297,
+      "step": 875
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989837306592567e-05,
+      "loss": 4.8876,
+      "step": 880
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989706642918291e-05,
+      "loss": 5.0282,
+      "step": 885
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989575145482583e-05,
+      "loss": 4.9422,
+      "step": 890
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989442814307422e-05,
+      "loss": 4.8476,
+      "step": 895
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989309649414914e-05,
+      "loss": 4.7497,
+      "step": 900
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989175650827314e-05,
+      "loss": 4.9203,
+      "step": 905
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.98904081856701e-05,
+      "loss": 4.9654,
+      "step": 910
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988905152656534e-05,
+      "loss": 4.9439,
+      "step": 915
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988768653118554e-05,
+      "loss": 4.9474,
+      "step": 920
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988631319975881e-05,
+      "loss": 4.8308,
+      "step": 925
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988493153251459e-05,
+      "loss": 4.936,
+      "step": 930
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988354152968379e-05,
+      "loss": 4.8146,
+      "step": 935
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988214319149865e-05,
+      "loss": 4.9064,
+      "step": 940
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988073651819281e-05,
+      "loss": 4.9894,
+      "step": 945
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987932151000137e-05,
+      "loss": 4.8355,
+      "step": 950
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987789816716073e-05,
+      "loss": 4.7528,
+      "step": 955
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987646648990876e-05,
+      "loss": 4.8677,
+      "step": 960
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.987502647848466e-05,
+      "loss": 4.8036,
+      "step": 965
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.987357813312905e-05,
+      "loss": 4.7445,
+      "step": 970
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.987212145408395e-05,
+      "loss": 4.8045,
+      "step": 975
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.987065644159278e-05,
+      "loss": 4.8328,
+      "step": 980
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.986918309590031e-05,
+      "loss": 4.8539,
+      "step": 985
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.986770141725277e-05,
+      "loss": 4.8898,
+      "step": 990
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.986621140589772e-05,
+      "loss": 4.9226,
+      "step": 995
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.986471306208412e-05,
+      "loss": 4.8012,
+      "step": 1000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.986320638606235e-05,
+      "loss": 4.8914,
+      "step": 1005
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.986169137808419e-05,
+      "loss": 4.646,
+      "step": 1010
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.986016803840275e-05,
+      "loss": 4.8681,
+      "step": 1015
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.985863636727262e-05,
+      "loss": 4.8381,
+      "step": 1020
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.985709636494971e-05,
+      "loss": 4.7454,
+      "step": 1025
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.985554803169134e-05,
+      "loss": 4.8566,
+      "step": 1030
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.985399136775627e-05,
+      "loss": 4.7265,
+      "step": 1035
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.985242637340458e-05,
+      "loss": 4.7943,
+      "step": 1040
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.985085304889778e-05,
+      "loss": 4.7365,
+      "step": 1045
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984927139449877e-05,
+      "loss": 4.7207,
+      "step": 1050
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984768141047184e-05,
+      "loss": 4.859,
+      "step": 1055
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984608309708266e-05,
+      "loss": 4.7324,
+      "step": 1060
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984447645459831e-05,
+      "loss": 4.9531,
+      "step": 1065
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984286148328728e-05,
+      "loss": 4.7689,
+      "step": 1070
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984123818341937e-05,
+      "loss": 4.8381,
+      "step": 1075
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983960655526587e-05,
+      "loss": 4.6922,
+      "step": 1080
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983796659909941e-05,
+      "loss": 4.827,
+      "step": 1085
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983631831519401e-05,
+      "loss": 4.5641,
+      "step": 1090
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983466170382511e-05,
+      "loss": 4.7204,
+      "step": 1095
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983299676526949e-05,
+      "loss": 4.6918,
+      "step": 1100
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.98313234998054e-05,
+      "loss": 4.7275,
+      "step": 1105
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.982964190771242e-05,
+      "loss": 4.8356,
+      "step": 1110
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.98279519892715e-05,
+      "loss": 4.708,
+      "step": 1115
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.982625374476508e-05,
+      "loss": 4.7869,
+      "step": 1120
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.98245471744769e-05,
+      "loss": 4.698,
+      "step": 1125
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.982283227869211e-05,
+      "loss": 4.8878,
+      "step": 1130
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.982110905769729e-05,
+      "loss": 4.8722,
+      "step": 1135
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.981937751178037e-05,
+      "loss": 4.6926,
+      "step": 1140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.981763764123067e-05,
+      "loss": 4.6914,
+      "step": 1145
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.981588944633893e-05,
+      "loss": 4.7965,
+      "step": 1150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.981413292739727e-05,
+      "loss": 4.8769,
+      "step": 1155
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.981236808469922e-05,
+      "loss": 4.7877,
+      "step": 1160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.981059491853964e-05,
+      "loss": 4.7068,
+      "step": 1165
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.980881342921482e-05,
+      "loss": 4.6711,
+      "step": 1170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.980702361702246e-05,
+      "loss": 5.1481,
+      "step": 1175
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.980522548226162e-05,
+      "loss": 4.7324,
+      "step": 1180
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.980341902523279e-05,
+      "loss": 4.6958,
+      "step": 1185
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.980160424623776e-05,
+      "loss": 4.7482,
+      "step": 1190
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.979978114557984e-05,
+      "loss": 4.6878,
+      "step": 1195
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.979794972356362e-05,
+      "loss": 4.7356,
+      "step": 1200
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.979610998049516e-05,
+      "loss": 4.732,
+      "step": 1205
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.979426191668182e-05,
+      "loss": 4.6581,
+      "step": 1210
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.979240553243246e-05,
+      "loss": 4.6731,
+      "step": 1215
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.979054082805726e-05,
+      "loss": 4.6375,
+      "step": 1220
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.978866780386777e-05,
+      "loss": 4.6351,
+      "step": 1225
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.978678646017701e-05,
+      "loss": 4.6412,
+      "step": 1230
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.978489679729934e-05,
+      "loss": 4.6057,
+      "step": 1235
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.978299881555048e-05,
+      "loss": 4.6331,
+      "step": 1240
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.97810925152476e-05,
+      "loss": 4.6184,
+      "step": 1245
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.977917789670925e-05,
+      "loss": 4.5721,
+      "step": 1250
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.977725496025535e-05,
+      "loss": 4.6374,
+      "step": 1255
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.977532370620718e-05,
+      "loss": 4.6428,
+      "step": 1260
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.977338413488748e-05,
+      "loss": 4.5927,
+      "step": 1265
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.977143624662035e-05,
+      "loss": 4.6058,
+      "step": 1270
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.976948004173125e-05,
+      "loss": 4.6322,
+      "step": 1275
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.976751552054706e-05,
+      "loss": 4.6317,
+      "step": 1280
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.976554268339607e-05,
+      "loss": 4.6729,
+      "step": 1285
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.97635615306079e-05,
+      "loss": 4.6425,
+      "step": 1290
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.976157206251361e-05,
+      "loss": 4.6688,
+      "step": 1295
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.975957427944563e-05,
+      "loss": 4.6506,
+      "step": 1300
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.975756818173778e-05,
+      "loss": 4.5674,
+      "step": 1305
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.975555376972529e-05,
+      "loss": 4.6175,
+      "step": 1310
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.975353104374474e-05,
+      "loss": 4.615,
+      "step": 1315
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.975150000413411e-05,
+      "loss": 4.5196,
+      "step": 1320
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.97494606512328e-05,
+      "loss": 4.5903,
+      "step": 1325
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.974741298538158e-05,
+      "loss": 4.5522,
+      "step": 1330
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.974535700692257e-05,
+      "loss": 4.524,
+      "step": 1335
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.974329271619938e-05,
+      "loss": 4.5628,
+      "step": 1340
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.97412201135569e-05,
+      "loss": 4.5312,
+      "step": 1345
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.973913919934147e-05,
+      "loss": 4.5413,
+      "step": 1350
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.973704997390079e-05,
+      "loss": 4.5405,
+      "step": 1355
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.973495243758398e-05,
+      "loss": 4.4892,
+      "step": 1360
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.973284659074151e-05,
+      "loss": 4.3883,
+      "step": 1365
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.973073243372528e-05,
+      "loss": 4.5406,
+      "step": 1370
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.972860996688853e-05,
+      "loss": 4.4999,
+      "step": 1375
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.972647919058595e-05,
+      "loss": 4.6377,
+      "step": 1380
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.972434010517358e-05,
+      "loss": 4.6257,
+      "step": 1385
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.972219271100882e-05,
+      "loss": 4.602,
+      "step": 1390
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.972003700845052e-05,
+      "loss": 4.6264,
+      "step": 1395
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.971787299785888e-05,
+      "loss": 4.6726,
+      "step": 1400
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.97157006795955e-05,
+      "loss": 4.5058,
+      "step": 1405
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.971352005402338e-05,
+      "loss": 4.6742,
+      "step": 1410
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.971133112150688e-05,
+      "loss": 4.6117,
+      "step": 1415
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.970913388241177e-05,
+      "loss": 4.5029,
+      "step": 1420
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.970692833710517e-05,
+      "loss": 4.523,
+      "step": 1425
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.970471448595566e-05,
+      "loss": 4.4897,
+      "step": 1430
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.970249232933315e-05,
+      "loss": 4.5523,
+      "step": 1435
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.970026186760896e-05,
+      "loss": 4.4867,
+      "step": 1440
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.969802310115577e-05,
+      "loss": 4.5104,
+      "step": 1445
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.96957760303477e-05,
+      "loss": 4.5177,
+      "step": 1450
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.969352065556022e-05,
+      "loss": 4.5617,
+      "step": 1455
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.969125697717017e-05,
+      "loss": 4.5853,
+      "step": 1460
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.968898499555582e-05,
+      "loss": 4.4209,
+      "step": 1465
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.968670471109682e-05,
+      "loss": 4.4264,
+      "step": 1470
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.968441612417418e-05,
+      "loss": 4.5204,
+      "step": 1475
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.968211923517032e-05,
+      "loss": 4.4371,
+      "step": 1480
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.967981404446905e-05,
+      "loss": 4.4349,
+      "step": 1485
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.967750055245555e-05,
+      "loss": 4.5116,
+      "step": 1490
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.967517875951638e-05,
+      "loss": 4.519,
+      "step": 1495
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.967284866603955e-05,
+      "loss": 4.4366,
+      "step": 1500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.967051027241436e-05,
+      "loss": 4.5136,
+      "step": 1505
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.966816357903155e-05,
+      "loss": 4.3133,
+      "step": 1510
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.96658085862833e-05,
+      "loss": 4.492,
+      "step": 1515
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.966344529456305e-05,
+      "loss": 4.5047,
+      "step": 1520
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.966107370426573e-05,
+      "loss": 4.4553,
+      "step": 1525
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.965869381578765e-05,
+      "loss": 4.5118,
+      "step": 1530
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.965630562952643e-05,
+      "loss": 4.4061,
+      "step": 1535
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.965390914588115e-05,
+      "loss": 4.4017,
+      "step": 1540
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.965150436525225e-05,
+      "loss": 4.5118,
+      "step": 1545
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.964909128804159e-05,
+      "loss": 4.4513,
+      "step": 1550
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.964666991465234e-05,
+      "loss": 4.5405,
+      "step": 1555
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.964424024548914e-05,
+      "loss": 4.4045,
+      "step": 1560
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.964180228095793e-05,
+      "loss": 4.4224,
+      "step": 1565
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.963935602146612e-05,
+      "loss": 4.4212,
+      "step": 1570
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.963690146742248e-05,
+      "loss": 4.4177,
+      "step": 1575
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.963443861923715e-05,
+      "loss": 4.4654,
+      "step": 1580
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.963196747732164e-05,
+      "loss": 4.518,
+      "step": 1585
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.96294880420889e-05,
+      "loss": 4.4395,
+      "step": 1590
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.96270003139532e-05,
+      "loss": 4.6269,
+      "step": 1595
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.962450429333026e-05,
+      "loss": 4.4465,
+      "step": 1600
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.962199998063715e-05,
+      "loss": 4.413,
+      "step": 1605
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.961948737629231e-05,
+      "loss": 4.4611,
+      "step": 1610
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.961696648071561e-05,
+      "loss": 4.4328,
+      "step": 1615
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.961443729432828e-05,
+      "loss": 4.3426,
+      "step": 1620
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.961189981755294e-05,
+      "loss": 4.4174,
+      "step": 1625
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.960935405081357e-05,
+      "loss": 4.3696,
+      "step": 1630
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.960679999453559e-05,
+      "loss": 4.3472,
+      "step": 1635
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.960423764914575e-05,
+      "loss": 4.3783,
+      "step": 1640
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.960166701507221e-05,
+      "loss": 4.3715,
+      "step": 1645
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.959908809274453e-05,
+      "loss": 4.3358,
+      "step": 1650
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.959650088259362e-05,
+      "loss": 4.3129,
+      "step": 1655
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.959390538505181e-05,
+      "loss": 4.3346,
+      "step": 1660
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.959130160055279e-05,
+      "loss": 4.4165,
+      "step": 1665
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.958868952953163e-05,
+      "loss": 4.3284,
+      "step": 1670
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.95860691724248e-05,
+      "loss": 4.3436,
+      "step": 1675
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.958344052967018e-05,
+      "loss": 4.4476,
+      "step": 1680
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.958080360170698e-05,
+      "loss": 4.3316,
+      "step": 1685
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.957815838897585e-05,
+      "loss": 4.2783,
+      "step": 1690
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.957550489191874e-05,
+      "loss": 4.3726,
+      "step": 1695
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.957284311097908e-05,
+      "loss": 4.3805,
+      "step": 1700
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.957017304660164e-05,
+      "loss": 4.3507,
+      "step": 1705
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.956749469923258e-05,
+      "loss": 4.3222,
+      "step": 1710
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.956480806931942e-05,
+      "loss": 4.2696,
+      "step": 1715
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.956211315731112e-05,
+      "loss": 4.4003,
+      "step": 1720
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.955940996365796e-05,
+      "loss": 4.4589,
+      "step": 1725
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.955669848881166e-05,
+      "loss": 4.3189,
+      "step": 1730
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.955397873322527e-05,
+      "loss": 4.4033,
+      "step": 1735
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.955125069735327e-05,
+      "loss": 4.5102,
+      "step": 1740
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.95485143816515e-05,
+      "loss": 4.3762,
+      "step": 1745
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.95457697865772e-05,
+      "loss": 4.3254,
+      "step": 1750
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.954301691258894e-05,
+      "loss": 4.3429,
+      "step": 1755
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.954025576014674e-05,
+      "loss": 4.3656,
+      "step": 1760
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.953748632971201e-05,
+      "loss": 4.356,
+      "step": 1765
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.953470862174748e-05,
+      "loss": 4.3829,
+      "step": 1770
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.953192263671728e-05,
+      "loss": 4.2554,
+      "step": 1775
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.952912837508697e-05,
+      "loss": 4.2784,
+      "step": 1780
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.952632583732343e-05,
+      "loss": 4.3586,
+      "step": 1785
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.952351502389498e-05,
+      "loss": 4.4031,
+      "step": 1790
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.95206959352713e-05,
+      "loss": 4.2831,
+      "step": 1795
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.951786857192341e-05,
+      "loss": 4.2545,
+      "step": 1800
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.951503293432378e-05,
+      "loss": 4.2606,
+      "step": 1805
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.951218902294625e-05,
+      "loss": 4.3101,
+      "step": 1810
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.9509336838266e-05,
+      "loss": 4.2869,
+      "step": 1815
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.950647638075963e-05,
+      "loss": 4.3369,
+      "step": 1820
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.950360765090511e-05,
+      "loss": 4.3647,
+      "step": 1825
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.950073064918179e-05,
+      "loss": 4.2848,
+      "step": 1830
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.949784537607042e-05,
+      "loss": 4.2729,
+      "step": 1835
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.949495183205311e-05,
+      "loss": 4.3402,
+      "step": 1840
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.949205001761334e-05,
+      "loss": 4.3609,
+      "step": 1845
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.948913993323603e-05,
+      "loss": 4.2556,
+      "step": 1850
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.948622157940744e-05,
+      "loss": 4.2856,
+      "step": 1855
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.948329495661517e-05,
+      "loss": 4.3539,
+      "step": 1860
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.94803600653483e-05,
+      "loss": 4.4524,
+      "step": 1865
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.947741690609722e-05,
+      "loss": 4.3196,
+      "step": 1870
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.94744654793537e-05,
+      "loss": 4.309,
+      "step": 1875
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.947150578561095e-05,
+      "loss": 4.255,
+      "step": 1880
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.946853782536351e-05,
+      "loss": 4.2917,
+      "step": 1885
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.946556159910732e-05,
+      "loss": 4.3717,
+      "step": 1890
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.946257710733966e-05,
+      "loss": 4.1936,
+      "step": 1895
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.945958435055927e-05,
+      "loss": 4.2946,
+      "step": 1900
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.945658332926622e-05,
+      "loss": 4.2894,
+      "step": 1905
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.945357404396197e-05,
+      "loss": 4.2701,
+      "step": 1910
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.945055649514934e-05,
+      "loss": 4.3815,
+      "step": 1915
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.944753068333256e-05,
+      "loss": 4.3336,
+      "step": 1920
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.944449660901725e-05,
+      "loss": 4.2318,
+      "step": 1925
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.944145427271036e-05,
+      "loss": 4.2905,
+      "step": 1930
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.94384036749203e-05,
+      "loss": 4.2754,
+      "step": 1935
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.943534481615677e-05,
+      "loss": 4.2598,
+      "step": 1940
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.943227769693091e-05,
+      "loss": 4.2305,
+      "step": 1945
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.942920231775524e-05,
+      "loss": 4.2706,
+      "step": 1950
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.942611867914363e-05,
+      "loss": 4.2538,
+      "step": 1955
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.942302678161133e-05,
+      "loss": 4.1425,
+      "step": 1960
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.9419926625675e-05,
+      "loss": 4.2438,
+      "step": 1965
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.941681821185265e-05,
+      "loss": 4.2277,
+      "step": 1970
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.941370154066371e-05,
+      "loss": 4.119,
+      "step": 1975
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.941057661262895e-05,
+      "loss": 4.2414,
+      "step": 1980
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.940744342827055e-05,
+      "loss": 4.2727,
+      "step": 1985
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.940430198811202e-05,
+      "loss": 4.0989,
+      "step": 1990
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.940115229267831e-05,
+      "loss": 4.0957,
+      "step": 1995
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.939799434249571e-05,
+      "loss": 4.1946,
+      "step": 2000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.93948281380919e-05,
+      "loss": 4.3229,
+      "step": 2005
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.939165367999597e-05,
+      "loss": 4.2958,
+      "step": 2010
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.938847096873831e-05,
+      "loss": 4.2983,
+      "step": 2015
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.938528000485078e-05,
+      "loss": 4.3452,
+      "step": 2020
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.938208078886655e-05,
+      "loss": 4.2009,
+      "step": 2025
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.937887332132023e-05,
+      "loss": 4.2044,
+      "step": 2030
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.937565760274776e-05,
+      "loss": 4.1105,
+      "step": 2035
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.937243363368645e-05,
+      "loss": 4.1222,
+      "step": 2040
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.936920141467505e-05,
+      "loss": 4.2173,
+      "step": 2045
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.936596094625364e-05,
+      "loss": 4.3473,
+      "step": 2050
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.936271222896367e-05,
+      "loss": 4.2364,
+      "step": 2055
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.935945526334803e-05,
+      "loss": 4.2413,
+      "step": 2060
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.935619004995089e-05,
+      "loss": 4.0635,
+      "step": 2065
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.935291658931791e-05,
+      "loss": 4.0875,
+      "step": 2070
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.934963488199604e-05,
+      "loss": 4.2043,
+      "step": 2075
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.934634492853366e-05,
+      "loss": 4.2005,
+      "step": 2080
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.93430467294805e-05,
+      "loss": 4.1492,
+      "step": 2085
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.933974028538768e-05,
+      "loss": 4.1076,
+      "step": 2090
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.933642559680769e-05,
+      "loss": 4.2242,
+      "step": 2095
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.93331026642944e-05,
+      "loss": 4.2202,
+      "step": 2100
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.932977148840307e-05,
+      "loss": 4.3332,
+      "step": 2105
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.93264320696903e-05,
+      "loss": 4.1289,
+      "step": 2110
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.932308440871413e-05,
+      "loss": 4.1883,
+      "step": 2115
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.931972850603394e-05,
+      "loss": 4.241,
+      "step": 2120
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.931636436221048e-05,
+      "loss": 4.1596,
+      "step": 2125
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.931299197780586e-05,
+      "loss": 4.1573,
+      "step": 2130
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.930961135338361e-05,
+      "loss": 4.0226,
+      "step": 2135
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.930622248950864e-05,
+      "loss": 4.1538,
+      "step": 2140
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.93028253867472e-05,
+      "loss": 4.147,
+      "step": 2145
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.929942004566695e-05,
+      "loss": 4.0987,
+      "step": 2150
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.929600646683686e-05,
+      "loss": 4.1673,
+      "step": 2155
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.929258465082738e-05,
+      "loss": 4.0596,
+      "step": 2160
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.928915459821027e-05,
+      "loss": 4.5049,
+      "step": 2165
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.928571630955865e-05,
+      "loss": 4.2884,
+      "step": 2170
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.928226978544708e-05,
+      "loss": 4.2276,
+      "step": 2175
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.927881502645146e-05,
+      "loss": 4.1696,
+      "step": 2180
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.927535203314904e-05,
+      "loss": 4.0503,
+      "step": 2185
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.927188080611849e-05,
+      "loss": 4.1275,
+      "step": 2190
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.926840134593984e-05,
+      "loss": 4.1775,
+      "step": 2195
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.92649136531945e-05,
+      "loss": 4.1554,
+      "step": 2200
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.926141772846525e-05,
+      "loss": 4.0437,
+      "step": 2205
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.925791357233623e-05,
+      "loss": 4.0998,
+      "step": 2210
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.925440118539298e-05,
+      "loss": 4.1554,
+      "step": 2215
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.925088056822241e-05,
+      "loss": 4.1006,
+      "step": 2220
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.924735172141281e-05,
+      "loss": 4.2335,
+      "step": 2225
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.924381464555381e-05,
+      "loss": 4.0443,
+      "step": 2230
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.924026934123647e-05,
+      "loss": 4.2927,
+      "step": 2235
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.92367158090532e-05,
+      "loss": 4.3142,
+      "step": 2240
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.923315404959775e-05,
+      "loss": 4.2492,
+      "step": 2245
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.92295840634653e-05,
+      "loss": 4.1996,
+      "step": 2250
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.922600585125237e-05,
+      "loss": 4.1545,
+      "step": 2255
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.922241941355688e-05,
+      "loss": 4.0398,
+      "step": 2260
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.921882475097811e-05,
+      "loss": 4.0101,
+      "step": 2265
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.921522186411671e-05,
+      "loss": 4.2693,
+      "step": 2270
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.92116107535747e-05,
+      "loss": 4.0601,
+      "step": 2275
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.920799141995549e-05,
+      "loss": 4.0236,
+      "step": 2280
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.920436386386387e-05,
+      "loss": 4.1213,
+      "step": 2285
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.920072808590596e-05,
+      "loss": 4.0883,
+      "step": 2290
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.91970840866893e-05,
+      "loss": 4.033,
+      "step": 2295
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.919343186682282e-05,
+      "loss": 4.0692,
+      "step": 2300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.918977142691675e-05,
+      "loss": 4.1784,
+      "step": 2305
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.918610276758273e-05,
+      "loss": 4.042,
+      "step": 2310
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.918242588943383e-05,
+      "loss": 3.9946,
+      "step": 2315
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.91787407930844e-05,
+      "loss": 3.9887,
+      "step": 2320
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.917504747915023e-05,
+      "loss": 4.0438,
+      "step": 2325
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.917134594824843e-05,
+      "loss": 4.0406,
+      "step": 2330
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.916763620099754e-05,
+      "loss": 4.1552,
+      "step": 2335
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.916391823801743e-05,
+      "loss": 4.1021,
+      "step": 2340
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.916019205992937e-05,
+      "loss": 4.0586,
+      "step": 2345
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.915645766735597e-05,
+      "loss": 3.9955,
+      "step": 2350
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.915271506092126e-05,
+      "loss": 4.0692,
+      "step": 2355
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.91489642412506e-05,
+      "loss": 3.9962,
+      "step": 2360
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.914520520897072e-05,
+      "loss": 4.097,
+      "step": 2365
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.914143796470978e-05,
+      "loss": 4.0168,
+      "step": 2370
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.913766250909726e-05,
+      "loss": 4.1221,
+      "step": 2375
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.9133878842764e-05,
+      "loss": 4.1171,
+      "step": 2380
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.913008696634226e-05,
+      "loss": 4.187,
+      "step": 2385
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.912628688046564e-05,
+      "loss": 4.0515,
+      "step": 2390
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.912247858576914e-05,
+      "loss": 4.0311,
+      "step": 2395
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.911866208288906e-05,
+      "loss": 4.1619,
+      "step": 2400
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.911483737246319e-05,
+      "loss": 4.1251,
+      "step": 2405
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.911100445513058e-05,
+      "loss": 4.02,
+      "step": 2410
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.910716333153169e-05,
+      "loss": 4.1844,
+      "step": 2415
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.910331400230841e-05,
+      "loss": 4.077,
+      "step": 2420
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.90994564681039e-05,
+      "loss": 4.1996,
+      "step": 2425
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.909559072956275e-05,
+      "loss": 4.0839,
+      "step": 2430
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.909171678733091e-05,
+      "loss": 4.0997,
+      "step": 2435
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.908783464205573e-05,
+      "loss": 3.9856,
+      "step": 2440
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.908394429438585e-05,
+      "loss": 4.0587,
+      "step": 2445
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.908004574497139e-05,
+      "loss": 4.1384,
+      "step": 2450
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.907613899446372e-05,
+      "loss": 4.0738,
+      "step": 2455
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.90722240435157e-05,
+      "loss": 4.0275,
+      "step": 2460
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.906830089278148e-05,
+      "loss": 4.0295,
+      "step": 2465
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.906436954291661e-05,
+      "loss": 4.0499,
+      "step": 2470
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.906042999457798e-05,
+      "loss": 4.0244,
+      "step": 2475
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.905648224842391e-05,
+      "loss": 4.0765,
+      "step": 2480
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.905252630511405e-05,
+      "loss": 4.0101,
+      "step": 2485
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.904856216530938e-05,
+      "loss": 4.0157,
+      "step": 2490
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.904458982967235e-05,
+      "loss": 3.8306,
+      "step": 2495
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.904060929886668e-05,
+      "loss": 4.0317,
+      "step": 2500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.903662057355751e-05,
+      "loss": 4.0475,
+      "step": 2505
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.903262365441137e-05,
+      "loss": 4.0576,
+      "step": 2510
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.90286185420961e-05,
+      "loss": 4.0157,
+      "step": 2515
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.902460523728094e-05,
+      "loss": 3.9789,
+      "step": 2520
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.902058374063651e-05,
+      "loss": 3.9531,
+      "step": 2525
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.90165540528348e-05,
+      "loss": 3.9932,
+      "step": 2530
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.901251617454913e-05,
+      "loss": 3.9898,
+      "step": 2535
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.900847010645422e-05,
+      "loss": 4.1109,
+      "step": 2540
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.900441584922616e-05,
+      "loss": 4.045,
+      "step": 2545
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.90003534035424e-05,
+      "loss": 4.024,
+      "step": 2550
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.899628277008175e-05,
+      "loss": 3.9861,
+      "step": 2555
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.899220394952441e-05,
+      "loss": 4.043,
+      "step": 2560
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.898811694255195e-05,
+      "loss": 3.9795,
+      "step": 2565
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.898402174984726e-05,
+      "loss": 3.9555,
+      "step": 2570
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.897991837209464e-05,
+      "loss": 4.0535,
+      "step": 2575
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.897580680997977e-05,
+      "loss": 3.9101,
+      "step": 2580
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.897168706418965e-05,
+      "loss": 3.9276,
+      "step": 2585
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.89675591354127e-05,
+      "loss": 3.9509,
+      "step": 2590
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.896342302433866e-05,
+      "loss": 3.9438,
+      "step": 2595
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.895927873165867e-05,
+      "loss": 4.0427,
+      "step": 2600
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.895512625806522e-05,
+      "loss": 3.9635,
+      "step": 2605
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.895096560425218e-05,
+      "loss": 3.9184,
+      "step": 2610
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.894679677091477e-05,
+      "loss": 3.9456,
+      "step": 2615
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.89426197587496e-05,
+      "loss": 3.9937,
+      "step": 2620
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.893843456845462e-05,
+      "loss": 3.9593,
+      "step": 2625
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.893424120072916e-05,
+      "loss": 3.9545,
+      "step": 2630
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.893003965627394e-05,
+      "loss": 3.9872,
+      "step": 2635
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.8925829935791e-05,
+      "loss": 4.08,
+      "step": 2640
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.892161203998375e-05,
+      "loss": 4.0257,
+      "step": 2645
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.891738596955703e-05,
+      "loss": 4.0275,
+      "step": 2650
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.891315172521696e-05,
+      "loss": 3.963,
+      "step": 2655
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.890890930767111e-05,
+      "loss": 4.0887,
+      "step": 2660
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.890465871762832e-05,
+      "loss": 3.9457,
+      "step": 2665
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.89003999557989e-05,
+      "loss": 3.9796,
+      "step": 2670
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.889613302289443e-05,
+      "loss": 4.0524,
+      "step": 2675
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.889185791962792e-05,
+      "loss": 3.9846,
+      "step": 2680
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.888757464671372e-05,
+      "loss": 4.0093,
+      "step": 2685
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.888328320486755e-05,
+      "loss": 3.9674,
+      "step": 2690
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.88789835948065e-05,
+      "loss": 3.9299,
+      "step": 2695
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.887467581724901e-05,
+      "loss": 4.0063,
+      "step": 2700
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.887035987291488e-05,
+      "loss": 3.8743,
+      "step": 2705
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.88660357625253e-05,
+      "loss": 3.8853,
+      "step": 2710
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.886170348680284e-05,
+      "loss": 3.9418,
+      "step": 2715
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.885736304647139e-05,
+      "loss": 3.8953,
+      "step": 2720
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.885301444225619e-05,
+      "loss": 3.9489,
+      "step": 2725
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.884865767488391e-05,
+      "loss": 3.9078,
+      "step": 2730
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.884429274508256e-05,
+      "loss": 3.9135,
+      "step": 2735
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.883991965358146e-05,
+      "loss": 3.9569,
+      "step": 2740
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.883553840111136e-05,
+      "loss": 3.8849,
+      "step": 2745
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.883114898840437e-05,
+      "loss": 3.9443,
+      "step": 2750
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.882675141619391e-05,
+      "loss": 3.9928,
+      "step": 2755
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.882234568521482e-05,
+      "loss": 3.837,
+      "step": 2760
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.881793179620328e-05,
+      "loss": 3.8162,
+      "step": 2765
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.881350974989682e-05,
+      "loss": 3.8586,
+      "step": 2770
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.880907954703437e-05,
+      "loss": 3.9087,
+      "step": 2775
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.880464118835619e-05,
+      "loss": 4.0122,
+      "step": 2780
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.880019467460391e-05,
+      "loss": 4.0206,
+      "step": 2785
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.879574000652052e-05,
+      "loss": 3.9516,
+      "step": 2790
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.87912771848504e-05,
+      "loss": 3.989,
+      "step": 2795
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.878680621033925e-05,
+      "loss": 3.8408,
+      "step": 2800
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.878232708373415e-05,
+      "loss": 3.9781,
+      "step": 2805
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.877783980578357e-05,
+      "loss": 3.8178,
+      "step": 2810
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.877334437723728e-05,
+      "loss": 3.8449,
+      "step": 2815
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.87688407988465e-05,
+      "loss": 3.8946,
+      "step": 2820
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.876432907136372e-05,
+      "loss": 3.897,
+      "step": 2825
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.875980919554287e-05,
+      "loss": 3.9302,
+      "step": 2830
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.875528117213915e-05,
+      "loss": 3.8802,
+      "step": 2835
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.875074500190922e-05,
+      "loss": 3.9745,
+      "step": 2840
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.874620068561104e-05,
+      "loss": 4.3549,
+      "step": 2845
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.874164822400396e-05,
+      "loss": 3.9927,
+      "step": 2850
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.873708761784866e-05,
+      "loss": 3.9262,
+      "step": 2855
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.873251886790722e-05,
+      "loss": 3.9385,
+      "step": 2860
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.872794197494307e-05,
+      "loss": 3.9425,
+      "step": 2865
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.872335693972094e-05,
+      "loss": 3.9135,
+      "step": 2870
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.871876376300702e-05,
+      "loss": 3.9211,
+      "step": 2875
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.871416244556881e-05,
+      "loss": 3.926,
+      "step": 2880
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.870955298817517e-05,
+      "loss": 3.9886,
+      "step": 2885
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.87049353915963e-05,
+      "loss": 4.0002,
+      "step": 2890
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.87003096566038e-05,
+      "loss": 3.9111,
+      "step": 2895
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.869567578397062e-05,
+      "loss": 3.8633,
+      "step": 2900
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.869103377447105e-05,
+      "loss": 3.8837,
+      "step": 2905
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.868638362888076e-05,
+      "loss": 3.9023,
+      "step": 2910
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.868172534797678e-05,
+      "loss": 3.9067,
+      "step": 2915
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.867705893253747e-05,
+      "loss": 3.9761,
+      "step": 2920
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.867238438334259e-05,
+      "loss": 3.953,
+      "step": 2925
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.866770170117324e-05,
+      "loss": 3.8207,
+      "step": 2930
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.866301088681185e-05,
+      "loss": 3.9826,
+      "step": 2935
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.865831194104227e-05,
+      "loss": 3.9058,
+      "step": 2940
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.865360486464967e-05,
+      "loss": 3.9326,
+      "step": 2945
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.864888965842056e-05,
+      "loss": 3.8834,
+      "step": 2950
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.864416632314287e-05,
+      "loss": 3.9193,
+      "step": 2955
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.863943485960583e-05,
+      "loss": 3.7672,
+      "step": 2960
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.863469526860005e-05,
+      "loss": 3.9402,
+      "step": 2965
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.862994755091752e-05,
+      "loss": 3.8603,
+      "step": 2970
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.862519170735154e-05,
+      "loss": 3.7597,
+      "step": 2975
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.86204277386968e-05,
+      "loss": 3.9385,
+      "step": 2980
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.861565564574935e-05,
+      "loss": 5.035,
+      "step": 2985
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.861087542930658e-05,
+      "loss": 4.0521,
+      "step": 2990
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.860608709016726e-05,
+      "loss": 3.9949,
+      "step": 2995
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.860129062913147e-05,
+      "loss": 4.0869,
+      "step": 3000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.859648604700072e-05,
+      "loss": 3.9137,
+      "step": 3005
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.859167334457781e-05,
+      "loss": 3.8948,
+      "step": 3010
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.858685252266696e-05,
+      "loss": 3.9508,
+      "step": 3015
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.858202358207367e-05,
+      "loss": 3.7666,
+      "step": 3020
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.857718652360486e-05,
+      "loss": 3.8961,
+      "step": 3025
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.857234134806878e-05,
+      "loss": 3.8693,
+      "step": 3030
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.856748805627503e-05,
+      "loss": 3.9589,
+      "step": 3035
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.856262664903462e-05,
+      "loss": 3.8615,
+      "step": 3040
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.855775712715981e-05,
+      "loss": 3.8059,
+      "step": 3045
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.855287949146432e-05,
+      "loss": 3.8478,
+      "step": 3050
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.854799374276318e-05,
+      "loss": 3.878,
+      "step": 3055
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.854309988187277e-05,
+      "loss": 3.9376,
+      "step": 3060
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.853819790961086e-05,
+      "loss": 3.8792,
+      "step": 3065
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.853328782679652e-05,
+      "loss": 3.918,
+      "step": 3070
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.852836963425021e-05,
+      "loss": 3.9282,
+      "step": 3075
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.852344333279375e-05,
+      "loss": 3.8889,
+      "step": 3080
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.851850892325032e-05,
+      "loss": 3.8811,
+      "step": 3085
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.851356640644443e-05,
+      "loss": 3.851,
+      "step": 3090
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.850861578320196e-05,
+      "loss": 3.8521,
+      "step": 3095
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.850365705435013e-05,
+      "loss": 3.8748,
+      "step": 3100
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.849869022071753e-05,
+      "loss": 3.7892,
+      "step": 3105
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.849371528313411e-05,
+      "loss": 3.7976,
+      "step": 3110
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.848873224243114e-05,
+      "loss": 3.7854,
+      "step": 3115
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.848374109944128e-05,
+      "loss": 3.8655,
+      "step": 3120
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.847874185499855e-05,
+      "loss": 3.837,
+      "step": 3125
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.847373450993828e-05,
+      "loss": 3.8447,
+      "step": 3130
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.846871906509719e-05,
+      "loss": 3.7887,
+      "step": 3135
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.846369552131334e-05,
+      "loss": 3.8066,
+      "step": 3140
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.845866387942614e-05,
+      "loss": 3.8397,
+      "step": 3145
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.845362414027637e-05,
+      "loss": 3.7737,
+      "step": 3150
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.844857630470614e-05,
+      "loss": 3.7712,
+      "step": 3155
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.844352037355893e-05,
+      "loss": 3.8574,
+      "step": 3160
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.843845634767957e-05,
+      "loss": 3.8046,
+      "step": 3165
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.843338422791424e-05,
+      "loss": 3.7648,
+      "step": 3170
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.842830401511048e-05,
+      "loss": 3.7875,
+      "step": 3175
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.842321571011714e-05,
+      "loss": 3.8488,
+      "step": 3180
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.841811931378448e-05,
+      "loss": 3.8221,
+      "step": 3185
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.841301482696413e-05,
+      "loss": 3.8852,
+      "step": 3190
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.840790225050896e-05,
+      "loss": 3.8047,
+      "step": 3195
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.840278158527328e-05,
+      "loss": 3.6973,
+      "step": 3200
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.839765283211278e-05,
+      "loss": 3.7926,
+      "step": 3205
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.83925159918844e-05,
+      "loss": 3.7088,
+      "step": 3210
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.838737106544651e-05,
+      "loss": 3.8835,
+      "step": 3215
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.838221805365881e-05,
+      "loss": 3.8466,
+      "step": 3220
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.837705695738236e-05,
+      "loss": 3.8245,
+      "step": 3225
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.837188777747954e-05,
+      "loss": 3.7622,
+      "step": 3230
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.836671051481411e-05,
+      "loss": 3.7452,
+      "step": 3235
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.836152517025116e-05,
+      "loss": 3.7525,
+      "step": 3240
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.835633174465715e-05,
+      "loss": 3.8598,
+      "step": 3245
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.83511302388999e-05,
+      "loss": 3.8884,
+      "step": 3250
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.834592065384853e-05,
+      "loss": 3.7904,
+      "step": 3255
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.834070299037357e-05,
+      "loss": 3.7095,
+      "step": 3260
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.833547724934687e-05,
+      "loss": 3.8553,
+      "step": 3265
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.833024343164159e-05,
+      "loss": 3.8422,
+      "step": 3270
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.832500153813234e-05,
+      "loss": 3.7534,
+      "step": 3275
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.831975156969498e-05,
+      "loss": 3.8553,
+      "step": 3280
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.831449352720677e-05,
+      "loss": 3.8863,
+      "step": 3285
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.830922741154631e-05,
+      "loss": 3.8166,
+      "step": 3290
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.830395322359357e-05,
+      "loss": 3.7816,
+      "step": 3295
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.829867096422983e-05,
+      "loss": 3.7348,
+      "step": 3300
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.829338063433773e-05,
+      "loss": 3.8275,
+      "step": 3305
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.828808223480126e-05,
+      "loss": 5.6449,
+      "step": 3310
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.828277576650578e-05,
+      "loss": 4.0402,
+      "step": 3315
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.827746123033796e-05,
+      "loss": 3.9687,
+      "step": 3320
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.827213862718587e-05,
+      "loss": 3.9168,
+      "step": 3325
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.826680795793887e-05,
+      "loss": 3.8513,
+      "step": 3330
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.826146922348773e-05,
+      "loss": 3.7937,
+      "step": 3335
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.825612242472448e-05,
+      "loss": 3.7999,
+      "step": 3340
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.82507675625426e-05,
+      "loss": 3.625,
+      "step": 3345
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.824540463783682e-05,
+      "loss": 3.7765,
+      "step": 3350
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.824003365150329e-05,
+      "loss": 3.8171,
+      "step": 3355
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.82346546044395e-05,
+      "loss": 3.7941,
+      "step": 3360
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.822926749754424e-05,
+      "loss": 3.9306,
+      "step": 3365
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.822387233171771e-05,
+      "loss": 3.7558,
+      "step": 3370
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.821846910786136e-05,
+      "loss": 3.8561,
+      "step": 3375
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.821305782687812e-05,
+      "loss": 3.7819,
+      "step": 3380
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.820763848967215e-05,
+      "loss": 3.8077,
+      "step": 3385
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.820221109714902e-05,
+      "loss": 3.7589,
+      "step": 3390
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.81967756502156e-05,
+      "loss": 3.7284,
+      "step": 3395
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.819133214978017e-05,
+      "loss": 3.7761,
+      "step": 3400
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.818588059675229e-05,
+      "loss": 3.8351,
+      "step": 3405
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.818042099204292e-05,
+      "loss": 3.6783,
+      "step": 3410
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.817495333656433e-05,
+      "loss": 3.809,
+      "step": 3415
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.816947763123013e-05,
+      "loss": 3.761,
+      "step": 3420
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.816399387695529e-05,
+      "loss": 3.8449,
+      "step": 3425
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.815850207465616e-05,
+      "loss": 3.8396,
+      "step": 3430
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.815300222525035e-05,
+      "loss": 3.7796,
+      "step": 3435
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.814749432965691e-05,
+      "loss": 3.8041,
+      "step": 3440
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.814197838879615e-05,
+      "loss": 3.6573,
+      "step": 3445
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.813645440358981e-05,
+      "loss": 3.6551,
+      "step": 3450
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.813092237496087e-05,
+      "loss": 3.7713,
+      "step": 3455
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.812538230383376e-05,
+      "loss": 3.7613,
+      "step": 3460
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.811983419113418e-05,
+      "loss": 3.7521,
+      "step": 3465
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.81142780377892e-05,
+      "loss": 3.6832,
+      "step": 3470
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.810871384472724e-05,
+      "loss": 3.7261,
+      "step": 3475
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.810314161287806e-05,
+      "loss": 3.8108,
+      "step": 3480
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.809756134317276e-05,
+      "loss": 3.7003,
+      "step": 3485
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.809197303654376e-05,
+      "loss": 3.7681,
+      "step": 3490
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.808637669392487e-05,
+      "loss": 3.847,
+      "step": 3495
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.808077231625119e-05,
+      "loss": 3.823,
+      "step": 3500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.807515990445922e-05,
+      "loss": 3.6917,
+      "step": 3505
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.806953945948676e-05,
+      "loss": 3.8537,
+      "step": 3510
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.806391098227297e-05,
+      "loss": 3.6382,
+      "step": 3515
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.805827447375835e-05,
+      "loss": 3.7406,
+      "step": 3520
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.805262993488473e-05,
+      "loss": 3.7847,
+      "step": 3525
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.804697736659532e-05,
+      "loss": 3.7608,
+      "step": 3530
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.804131676983458e-05,
+      "loss": 3.7161,
+      "step": 3535
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.803564814554844e-05,
+      "loss": 3.7841,
+      "step": 3540
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.802997149468408e-05,
+      "loss": 3.7847,
+      "step": 3545
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.802428681819003e-05,
+      "loss": 3.7839,
+      "step": 3550
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.801859411701621e-05,
+      "loss": 3.8298,
+      "step": 3555
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.801289339211384e-05,
+      "loss": 3.7002,
+      "step": 3560
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.800718464443547e-05,
+      "loss": 3.8393,
+      "step": 3565
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.800146787493503e-05,
+      "loss": 3.7462,
+      "step": 3570
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.799574308456775e-05,
+      "loss": 3.7076,
+      "step": 3575
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.799001027429026e-05,
+      "loss": 3.673,
+      "step": 3580
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.798426944506048e-05,
+      "loss": 3.7247,
+      "step": 3585
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.797852059783766e-05,
+      "loss": 3.8097,
+      "step": 3590
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.79727637335824e-05,
+      "loss": 3.8171,
+      "step": 3595
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.79669988532567e-05,
+      "loss": 3.664,
+      "step": 3600
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.79612259578238e-05,
+      "loss": 3.7113,
+      "step": 3605
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.795544504824836e-05,
+      "loss": 3.7782,
+      "step": 3610
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.794965612549635e-05,
+      "loss": 3.6829,
+      "step": 3615
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.794385919053506e-05,
+      "loss": 3.6277,
+      "step": 3620
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.793805424433313e-05,
+      "loss": 3.7976,
+      "step": 3625
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.793224128786058e-05,
+      "loss": 3.7548,
+      "step": 3630
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.79264203220887e-05,
+      "loss": 3.745,
+      "step": 3635
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.792059134799016e-05,
+      "loss": 3.7491,
+      "step": 3640
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.791475436653896e-05,
+      "loss": 3.6839,
+      "step": 3645
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.790890937871046e-05,
+      "loss": 3.739,
+      "step": 3650
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.79030563854813e-05,
+      "loss": 3.6978,
+      "step": 3655
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.789719538782952e-05,
+      "loss": 3.6616,
+      "step": 3660
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.789132638673445e-05,
+      "loss": 3.8213,
+      "step": 3665
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.78854493831768e-05,
+      "loss": 3.7588,
+      "step": 3670
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.787956437813859e-05,
+      "loss": 3.673,
+      "step": 3675
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.787367137260317e-05,
+      "loss": 3.7372,
+      "step": 3680
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.786777036755525e-05,
+      "loss": 3.735,
+      "step": 3685
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.786186136398085e-05,
+      "loss": 3.8197,
+      "step": 3690
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.785594436286738e-05,
+      "loss": 3.697,
+      "step": 3695
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.78500193652035e-05,
+      "loss": 3.6629,
+      "step": 3700
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.784408637197929e-05,
+      "loss": 3.7381,
+      "step": 3705
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.783814538418612e-05,
+      "loss": 3.707,
+      "step": 3710
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.783219640281671e-05,
+      "loss": 3.7697,
+      "step": 3715
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.78262394288651e-05,
+      "loss": 3.6713,
+      "step": 3720
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.782027446332671e-05,
+      "loss": 3.7466,
+      "step": 3725
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.781430150719822e-05,
+      "loss": 3.758,
+      "step": 3730
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.780832056147772e-05,
+      "loss": 3.6797,
+      "step": 3735
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.78023316271646e-05,
+      "loss": 3.7399,
+      "step": 3740
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.779633470525957e-05,
+      "loss": 3.6649,
+      "step": 3745
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.77903297967647e-05,
+      "loss": 3.8,
+      "step": 3750
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.778431690268341e-05,
+      "loss": 3.7453,
+      "step": 3755
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.777829602402043e-05,
+      "loss": 3.6814,
+      "step": 3760
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.777226716178178e-05,
+      "loss": 3.7524,
+      "step": 3765
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.776623031697492e-05,
+      "loss": 3.7716,
+      "step": 3770
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.776018549060855e-05,
+      "loss": 3.6288,
+      "step": 3775
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.775413268369274e-05,
+      "loss": 3.6927,
+      "step": 3780
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.77480718972389e-05,
+      "loss": 3.6137,
+      "step": 3785
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.774200313225976e-05,
+      "loss": 3.6151,
+      "step": 3790
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.773592638976939e-05,
+      "loss": 3.7603,
+      "step": 3795
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.77298416707832e-05,
+      "loss": 3.691,
+      "step": 3800
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.772374897631791e-05,
+      "loss": 3.6866,
+      "step": 3805
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.771764830739159e-05,
+      "loss": 3.6389,
+      "step": 3810
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.771153966502365e-05,
+      "loss": 3.6684,
+      "step": 3815
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.77054230502348e-05,
+      "loss": 3.637,
+      "step": 3820
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.769929846404711e-05,
+      "loss": 3.7296,
+      "step": 3825
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.769316590748401e-05,
+      "loss": 3.7935,
+      "step": 3830
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.768702538157018e-05,
+      "loss": 3.8595,
+      "step": 3835
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.76808768873317e-05,
+      "loss": 3.73,
+      "step": 3840
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.767472042579595e-05,
+      "loss": 3.743,
+      "step": 3845
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.766855599799168e-05,
+      "loss": 3.7064,
+      "step": 3850
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.76623836049489e-05,
+      "loss": 3.6676,
+      "step": 3855
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.765620324769903e-05,
+      "loss": 3.7137,
+      "step": 3860
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.765001492727476e-05,
+      "loss": 3.7521,
+      "step": 3865
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.764381864471017e-05,
+      "loss": 3.7824,
+      "step": 3870
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.76376144010406e-05,
+      "loss": 3.592,
+      "step": 3875
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.763140219730278e-05,
+      "loss": 3.7791,
+      "step": 3880
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.762518203453472e-05,
+      "loss": 3.722,
+      "step": 3885
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.761895391377581e-05,
+      "loss": 3.7437,
+      "step": 3890
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.761271783606674e-05,
+      "loss": 3.7116,
+      "step": 3895
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.760647380244952e-05,
+      "loss": 3.7379,
+      "step": 3900
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.760022181396753e-05,
+      "loss": 3.5621,
+      "step": 3905
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.759396187166543e-05,
+      "loss": 3.7264,
+      "step": 3910
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.758769397658926e-05,
+      "loss": 3.5987,
+      "step": 3915
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.758141812978634e-05,
+      "loss": 3.638,
+      "step": 3920
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.757513433230534e-05,
+      "loss": 3.6802,
+      "step": 3925
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.756884258519627e-05,
+      "loss": 3.6041,
+      "step": 3930
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.756254288951046e-05,
+      "loss": 3.6221,
+      "step": 3935
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.755623524630055e-05,
+      "loss": 3.5305,
+      "step": 3940
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.754991965662055e-05,
+      "loss": 3.7603,
+      "step": 3945
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.754359612152574e-05,
+      "loss": 3.8047,
+      "step": 3950
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.753726464207278e-05,
+      "loss": 3.7631,
+      "step": 3955
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.753092521931963e-05,
+      "loss": 3.6675,
+      "step": 3960
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.752457785432558e-05,
+      "loss": 3.7345,
+      "step": 3965
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.751822254815126e-05,
+      "loss": 3.7224,
+      "step": 3970
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.751185930185861e-05,
+      "loss": 3.676,
+      "step": 3975
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.750548811651091e-05,
+      "loss": 3.5522,
+      "step": 3980
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.749910899317276e-05,
+      "loss": 3.7123,
+      "step": 3985
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.74927219329101e-05,
+      "loss": 3.8403,
+      "step": 3990
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.748632693679016e-05,
+      "loss": 3.7123,
+      "step": 3995
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.747992400588153e-05,
+      "loss": 3.6936,
+      "step": 4000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.747351314125413e-05,
+      "loss": 3.6473,
+      "step": 4005
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.746709434397916e-05,
+      "loss": 3.762,
+      "step": 4010
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.746066761512921e-05,
+      "loss": 3.6328,
+      "step": 4015
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.745423295577815e-05,
+      "loss": 3.581,
+      "step": 4020
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.744779036700118e-05,
+      "loss": 3.684,
+      "step": 4025
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.744133984987485e-05,
+      "loss": 3.6194,
+      "step": 4030
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.743488140547701e-05,
+      "loss": 3.6535,
+      "step": 4035
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.742841503488682e-05,
+      "loss": 3.9479,
+      "step": 4040
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.742194073918482e-05,
+      "loss": 3.7397,
+      "step": 4045
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.741545851945282e-05,
+      "loss": 3.6363,
+      "step": 4050
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.740896837677398e-05,
+      "loss": 3.642,
+      "step": 4055
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.740247031223279e-05,
+      "loss": 3.6875,
+      "step": 4060
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.739596432691502e-05,
+      "loss": 3.8143,
+      "step": 4065
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.738945042190785e-05,
+      "loss": 3.7233,
+      "step": 4070
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.738292859829968e-05,
+      "loss": 3.7717,
+      "step": 4075
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.73763988571803e-05,
+      "loss": 3.6477,
+      "step": 4080
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.73698611996408e-05,
+      "loss": 3.7512,
+      "step": 4085
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.736331562677361e-05,
+      "loss": 3.6476,
+      "step": 4090
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.735676213967247e-05,
+      "loss": 3.6673,
+      "step": 4095
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.735020073943244e-05,
+      "loss": 3.752,
+      "step": 4100
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.734363142714988e-05,
+      "loss": 3.7327,
+      "step": 4105
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.733705420392254e-05,
+      "loss": 3.6952,
+      "step": 4110
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.733046907084944e-05,
+      "loss": 3.7524,
+      "step": 4115
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.732387602903091e-05,
+      "loss": 3.5676,
+      "step": 4120
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.731727507956862e-05,
+      "loss": 3.6952,
+      "step": 4125
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.73106662235656e-05,
+      "loss": 3.6257,
+      "step": 4130
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.730404946212614e-05,
+      "loss": 3.6132,
+      "step": 4135
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.729742479635589e-05,
+      "loss": 3.5718,
+      "step": 4140
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.729079222736177e-05,
+      "loss": 3.6086,
+      "step": 4145
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.728415175625211e-05,
+      "loss": 3.6236,
+      "step": 4150
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.727750338413647e-05,
+      "loss": 3.5019,
+      "step": 4155
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.727084711212579e-05,
+      "loss": 3.6137,
+      "step": 4160
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.726418294133229e-05,
+      "loss": 3.7199,
+      "step": 4165
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.725751087286954e-05,
+      "loss": 3.6532,
+      "step": 4170
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.725083090785242e-05,
+      "loss": 3.6559,
+      "step": 4175
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.724414304739712e-05,
+      "loss": 3.6622,
+      "step": 4180
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.723744729262116e-05,
+      "loss": 3.6476,
+      "step": 4185
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.723074364464337e-05,
+      "loss": 3.7505,
+      "step": 4190
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.722403210458393e-05,
+      "loss": 3.6634,
+      "step": 4195
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.721731267356426e-05,
+      "loss": 3.6283,
+      "step": 4200
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.721058535270722e-05,
+      "loss": 3.6574,
+      "step": 4205
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.720385014313686e-05,
+      "loss": 3.6865,
+      "step": 4210
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.719710704597865e-05,
+      "loss": 3.6863,
+      "step": 4215
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.719035606235931e-05,
+      "loss": 3.6772,
+      "step": 4220
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.718359719340693e-05,
+      "loss": 3.5851,
+      "step": 4225
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.717683044025086e-05,
+      "loss": 3.688,
+      "step": 4230
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.717005580402184e-05,
+      "loss": 3.6158,
+      "step": 4235
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.716327328585183e-05,
+      "loss": 3.6761,
+      "step": 4240
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.715648288687421e-05,
+      "loss": 3.763,
+      "step": 4245
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.714968460822363e-05,
+      "loss": 3.6096,
+      "step": 4250
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.714287845103602e-05,
+      "loss": 3.6687,
+      "step": 4255
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.71360644164487e-05,
+      "loss": 3.6412,
+      "step": 4260
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.712924250560025e-05,
+      "loss": 3.63,
+      "step": 4265
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.712241271963059e-05,
+      "loss": 3.7477,
+      "step": 4270
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.711557505968097e-05,
+      "loss": 3.6353,
+      "step": 4275
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.710872952689391e-05,
+      "loss": 3.6508,
+      "step": 4280
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.71018761224133e-05,
+      "loss": 3.5869,
+      "step": 4285
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.709501484738429e-05,
+      "loss": 3.5786,
+      "step": 4290
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.70881457029534e-05,
+      "loss": 3.647,
+      "step": 4295
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.708126869026841e-05,
+      "loss": 3.6309,
+      "step": 4300
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.707438381047846e-05,
+      "loss": 3.6517,
+      "step": 4305
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.7067491064734e-05,
+      "loss": 3.7013,
+      "step": 4310
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.706059045418676e-05,
+      "loss": 3.6957,
+      "step": 4315
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.705368197998982e-05,
+      "loss": 3.6263,
+      "step": 4320
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.704676564329756e-05,
+      "loss": 3.6249,
+      "step": 4325
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.703984144526565e-05,
+      "loss": 3.6696,
+      "step": 4330
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.703290938705114e-05,
+      "loss": 3.7608,
+      "step": 4335
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.702596946981232e-05,
+      "loss": 3.6724,
+      "step": 4340
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.701902169470883e-05,
+      "loss": 3.6561,
+      "step": 4345
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.701206606290164e-05,
+      "loss": 3.6368,
+      "step": 4350
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.700510257555295e-05,
+      "loss": 3.6092,
+      "step": 4355
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.69981312338264e-05,
+      "loss": 3.5802,
+      "step": 4360
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.699115203888686e-05,
+      "loss": 3.7513,
+      "step": 4365
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.69841649919005e-05,
+      "loss": 3.6787,
+      "step": 4370
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.697717009403485e-05,
+      "loss": 3.6223,
+      "step": 4375
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.697016734645873e-05,
+      "loss": 3.6753,
+      "step": 4380
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.696315675034227e-05,
+      "loss": 3.5868,
+      "step": 4385
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.69561383068569e-05,
+      "loss": 3.5782,
+      "step": 4390
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.694911201717541e-05,
+      "loss": 3.6187,
+      "step": 4395
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.694207788247186e-05,
+      "loss": 3.632,
+      "step": 4400
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.693503590392159e-05,
+      "loss": 3.7813,
+      "step": 4405
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.692798608270132e-05,
+      "loss": 3.6642,
+      "step": 4410
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.692092841998904e-05,
+      "loss": 3.6332,
+      "step": 4415
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.691386291696407e-05,
+      "loss": 3.5621,
+      "step": 4420
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.690678957480702e-05,
+      "loss": 3.5346,
+      "step": 4425
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.689970839469981e-05,
+      "loss": 3.5651,
+      "step": 4430
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.689261937782571e-05,
+      "loss": 3.6751,
+      "step": 4435
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.688552252536924e-05,
+      "loss": 3.6088,
+      "step": 4440
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.687841783851626e-05,
+      "loss": 3.6234,
+      "step": 4445
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.687130531845394e-05,
+      "loss": 3.6421,
+      "step": 4450
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.686418496637076e-05,
+      "loss": 3.5713,
+      "step": 4455
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.685705678345652e-05,
+      "loss": 3.5946,
+      "step": 4460
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.684992077090227e-05,
+      "loss": 3.6672,
+      "step": 4465
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.684277692990045e-05,
+      "loss": 3.5773,
+      "step": 4470
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.683562526164475e-05,
+      "loss": 3.6287,
+      "step": 4475
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.68284657673302e-05,
+      "loss": 3.6571,
+      "step": 4480
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.682129844815311e-05,
+      "loss": 3.5638,
+      "step": 4485
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.681412330531112e-05,
+      "loss": 3.5653,
+      "step": 4490
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.680694034000318e-05,
+      "loss": 3.6672,
+      "step": 4495
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.67997495534295e-05,
+      "loss": 3.6525,
+      "step": 4500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.679255094679169e-05,
+      "loss": 3.6849,
+      "step": 4505
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.678534452129256e-05,
+      "loss": 3.7007,
+      "step": 4510
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.677813027813632e-05,
+      "loss": 3.5693,
+      "step": 4515
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.677090821852842e-05,
+      "loss": 3.6032,
+      "step": 4520
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.676367834367562e-05,
+      "loss": 3.6652,
+      "step": 4525
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.675644065478602e-05,
+      "loss": 3.4914,
+      "step": 4530
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.674919515306904e-05,
+      "loss": 3.6281,
+      "step": 4535
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.674194183973533e-05,
+      "loss": 3.6267,
+      "step": 4540
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.673468071599694e-05,
+      "loss": 3.6065,
+      "step": 4545
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.672741178306713e-05,
+      "loss": 3.6524,
+      "step": 4550
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.672013504216054e-05,
+      "loss": 3.5861,
+      "step": 4555
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.671285049449307e-05,
+      "loss": 3.6428,
+      "step": 4560
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.670555814128197e-05,
+      "loss": 3.6204,
+      "step": 4565
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.669825798374574e-05,
+      "loss": 3.6147,
+      "step": 4570
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.66909500231042e-05,
+      "loss": 3.6555,
+      "step": 4575
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.668363426057852e-05,
+      "loss": 3.5526,
+      "step": 4580
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.66763106973911e-05,
+      "loss": 3.5475,
+      "step": 4585
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.666897933476568e-05,
+      "loss": 3.6244,
+      "step": 4590
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.666164017392734e-05,
+      "loss": 3.6211,
+      "step": 4595
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.665429321610239e-05,
+      "loss": 3.5067,
+      "step": 4600
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.66469384625185e-05,
+      "loss": 3.5186,
+      "step": 4605
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.66395759144046e-05,
+      "loss": 3.5178,
+      "step": 4610
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.6632205572991e-05,
+      "loss": 3.6319,
+      "step": 4615
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.662482743950918e-05,
+      "loss": 3.545,
+      "step": 4620
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.661744151519206e-05,
+      "loss": 3.5132,
+      "step": 4625
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.661004780127373e-05,
+      "loss": 3.5144,
+      "step": 4630
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.660264629898974e-05,
+      "loss": 3.5842,
+      "step": 4635
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.65952370095768e-05,
+      "loss": 3.6219,
+      "step": 4640
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.6587819934273e-05,
+      "loss": 3.6019,
+      "step": 4645
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.658039507431768e-05,
+      "loss": 3.5206,
+      "step": 4650
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.657296243095151e-05,
+      "loss": 3.5996,
+      "step": 4655
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.656552200541648e-05,
+      "loss": 3.5706,
+      "step": 4660
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.655807379895583e-05,
+      "loss": 3.5521,
+      "step": 4665
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.655061781281413e-05,
+      "loss": 3.5344,
+      "step": 4670
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.654315404823729e-05,
+      "loss": 3.6347,
+      "step": 4675
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.653568250647242e-05,
+      "loss": 3.5441,
+      "step": 4680
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.652820318876802e-05,
+      "loss": 3.6726,
+      "step": 4685
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.652071609637385e-05,
+      "loss": 3.6549,
+      "step": 4690
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.651322123054097e-05,
+      "loss": 3.5699,
+      "step": 4695
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.650571859252175e-05,
+      "loss": 3.6689,
+      "step": 4700
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.649820818356985e-05,
+      "loss": 3.6058,
+      "step": 4705
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.649069000494025e-05,
+      "loss": 3.6588,
+      "step": 4710
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.648316405788916e-05,
+      "loss": 3.6361,
+      "step": 4715
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.647563034367419e-05,
+      "loss": 3.5272,
+      "step": 4720
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.646808886355417e-05,
+      "loss": 3.5258,
+      "step": 4725
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.646053961878925e-05,
+      "loss": 3.5737,
+      "step": 4730
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.64529826106409e-05,
+      "loss": 3.5678,
+      "step": 4735
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.644541784037186e-05,
+      "loss": 3.6815,
+      "step": 4740
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.643784530924616e-05,
+      "loss": 3.5651,
+      "step": 4745
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.643026501852916e-05,
+      "loss": 3.6035,
+      "step": 4750
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.64226769694875e-05,
+      "loss": 3.6101,
+      "step": 4755
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.641508116338908e-05,
+      "loss": 3.5237,
+      "step": 4760
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.640747760150319e-05,
+      "loss": 3.5055,
+      "step": 4765
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.639986628510032e-05,
+      "loss": 3.5372,
+      "step": 4770
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.639224721545229e-05,
+      "loss": 3.5466,
+      "step": 4775
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.638462039383223e-05,
+      "loss": 3.4762,
+      "step": 4780
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.637698582151454e-05,
+      "loss": 3.5713,
+      "step": 4785
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.636934349977493e-05,
+      "loss": 3.6026,
+      "step": 4790
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.636169342989043e-05,
+      "loss": 3.5484,
+      "step": 4795
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.635403561313932e-05,
+      "loss": 3.5494,
+      "step": 4800
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.634637005080119e-05,
+      "loss": 3.5182,
+      "step": 4805
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.633869674415693e-05,
+      "loss": 3.6614,
+      "step": 4810
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.633101569448873e-05,
+      "loss": 3.4653,
+      "step": 4815
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.632332690308004e-05,
+      "loss": 3.557,
+      "step": 4820
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.631563037121565e-05,
+      "loss": 3.5162,
+      "step": 4825
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.63079261001816e-05,
+      "loss": 3.469,
+      "step": 4830
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.630021409126529e-05,
+      "loss": 3.5431,
+      "step": 4835
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.629249434575531e-05,
+      "loss": 3.6473,
+      "step": 4840
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.628476686494164e-05,
+      "loss": 3.5992,
+      "step": 4845
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.627703165011551e-05,
+      "loss": 3.6098,
+      "step": 4850
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.626928870256944e-05,
+      "loss": 3.5712,
+      "step": 4855
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.626153802359724e-05,
+      "loss": 3.6521,
+      "step": 4860
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.625377961449402e-05,
+      "loss": 3.6194,
+      "step": 4865
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.62460134765562e-05,
+      "loss": 3.5589,
+      "step": 4870
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.623823961108145e-05,
+      "loss": 3.4931,
+      "step": 4875
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.623045801936876e-05,
+      "loss": 3.5199,
+      "step": 4880
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.622266870271842e-05,
+      "loss": 4.2079,
+      "step": 4885
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.6214871662432e-05,
+      "loss": 3.6408,
+      "step": 4890
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.620706689981232e-05,
+      "loss": 3.6619,
+      "step": 4895
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.619925441616354e-05,
+      "loss": 3.6512,
+      "step": 4900
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.619143421279112e-05,
+      "loss": 3.6706,
+      "step": 4905
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.618360629100176e-05,
+      "loss": 3.4855,
+      "step": 4910
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.61757706521035e-05,
+      "loss": 3.5348,
+      "step": 4915
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.616792729740564e-05,
+      "loss": 3.4652,
+      "step": 4920
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.616007622821877e-05,
+      "loss": 3.5961,
+      "step": 4925
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.615221744585476e-05,
+      "loss": 3.5874,
+      "step": 4930
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.614435095162679e-05,
+      "loss": 3.5356,
+      "step": 4935
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.613647674684936e-05,
+      "loss": 3.5186,
+      "step": 4940
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.612859483283818e-05,
+      "loss": 3.496,
+      "step": 4945
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.612070521091029e-05,
+      "loss": 3.6983,
+      "step": 4950
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.611280788238403e-05,
+      "loss": 3.6801,
+      "step": 4955
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.610490284857901e-05,
+      "loss": 3.5112,
+      "step": 4960
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.609699011081613e-05,
+      "loss": 3.5792,
+      "step": 4965
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.608906967041757e-05,
+      "loss": 3.6444,
+      "step": 4970
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.608114152870684e-05,
+      "loss": 3.4417,
+      "step": 4975
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.607320568700866e-05,
+      "loss": 3.5427,
+      "step": 4980
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.606526214664911e-05,
+      "loss": 3.5068,
+      "step": 4985
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.605731090895551e-05,
+      "loss": 3.6275,
+      "step": 4990
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.604935197525649e-05,
+      "loss": 3.5931,
+      "step": 4995
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.604138534688196e-05,
+      "loss": 3.5733,
+      "step": 5000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.60334110251631e-05,
+      "loss": 3.5268,
+      "step": 5005
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.602542901143241e-05,
+      "loss": 3.5471,
+      "step": 5010
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.601743930702363e-05,
+      "loss": 3.5555,
+      "step": 5015
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.600944191327184e-05,
+      "loss": 3.6071,
+      "step": 5020
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.600143683151335e-05,
+      "loss": 3.5473,
+      "step": 5025
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.59934240630858e-05,
+      "loss": 3.6309,
+      "step": 5030
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.598540360932806e-05,
+      "loss": 3.6021,
+      "step": 5035
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.597737547158035e-05,
+      "loss": 3.6231,
+      "step": 5040
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.596933965118413e-05,
+      "loss": 3.4823,
+      "step": 5045
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.596129614948217e-05,
+      "loss": 3.5626,
+      "step": 5050
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.595324496781849e-05,
+      "loss": 3.5059,
+      "step": 5055
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.594518610753841e-05,
+      "loss": 3.4634,
+      "step": 5060
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.593711956998855e-05,
+      "loss": 3.4374,
+      "step": 5065
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.59290453565168e-05,
+      "loss": 3.5191,
+      "step": 5070
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.592096346847232e-05,
+      "loss": 3.5191,
+      "step": 5075
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.591287390720555e-05,
+      "loss": 3.5593,
+      "step": 5080
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.590477667406828e-05,
+      "loss": 3.6117,
+      "step": 5085
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.589667177041347e-05,
+      "loss": 3.6147,
+      "step": 5090
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.588855919759544e-05,
+      "loss": 3.6184,
+      "step": 5095
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.588043895696978e-05,
+      "loss": 3.4713,
+      "step": 5100
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.587231104989334e-05,
+      "loss": 3.5528,
+      "step": 5105
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.586417547772426e-05,
+      "loss": 3.5017,
+      "step": 5110
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.585603224182196e-05,
+      "loss": 3.6074,
+      "step": 5115
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.584788134354718e-05,
+      "loss": 3.5423,
+      "step": 5120
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.583972278426186e-05,
+      "loss": 3.5479,
+      "step": 5125
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.583155656532929e-05,
+      "loss": 3.5738,
+      "step": 5130
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.582338268811402e-05,
+      "loss": 3.5586,
+      "step": 5135
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.581520115398185e-05,
+      "loss": 3.5753,
+      "step": 5140
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.580701196429991e-05,
+      "loss": 3.4486,
+      "step": 5145
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.579881512043655e-05,
+      "loss": 3.3608,
+      "step": 5150
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.579061062376146e-05,
+      "loss": 3.4878,
+      "step": 5155
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.578239847564559e-05,
+      "loss": 3.4722,
+      "step": 5160
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.577417867746113e-05,
+      "loss": 3.6016,
+      "step": 5165
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.576595123058158e-05,
+      "loss": 3.4319,
+      "step": 5170
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.575771613638175e-05,
+      "loss": 3.6129,
+      "step": 5175
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.574947339623765e-05,
+      "loss": 3.512,
+      "step": 5180
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.574122301152665e-05,
+      "loss": 3.5504,
+      "step": 5185
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.573296498362733e-05,
+      "loss": 3.4862,
+      "step": 5190
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.572469931391959e-05,
+      "loss": 3.4609,
+      "step": 5195
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.571642600378459e-05,
+      "loss": 3.5689,
+      "step": 5200
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.570814505460475e-05,
+      "loss": 3.545,
+      "step": 5205
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.569985646776382e-05,
+      "loss": 3.4409,
+      "step": 5210
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.569156024464677e-05,
+      "loss": 3.5387,
+      "step": 5215
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.568325638663989e-05,
+      "loss": 3.4474,
+      "step": 5220
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.567494489513068e-05,
+      "loss": 3.5845,
+      "step": 5225
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.5666625771508e-05,
+      "loss": 3.5753,
+      "step": 5230
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.565829901716194e-05,
+      "loss": 3.4789,
+      "step": 5235
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.564996463348384e-05,
+      "loss": 3.6066,
+      "step": 5240
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.564162262186638e-05,
+      "loss": 3.6042,
+      "step": 5245
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.563327298370347e-05,
+      "loss": 3.5206,
+      "step": 5250
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.56249157203903e-05,
+      "loss": 3.5466,
+      "step": 5255
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.561655083332333e-05,
+      "loss": 3.533,
+      "step": 5260
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.56081783239003e-05,
+      "loss": 3.5653,
+      "step": 5265
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.559979819352024e-05,
+      "loss": 3.4425,
+      "step": 5270
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.559141044358342e-05,
+      "loss": 3.5046,
+      "step": 5275
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.558301507549143e-05,
+      "loss": 3.6096,
+      "step": 5280
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.557461209064708e-05,
+      "loss": 3.454,
+      "step": 5285
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.556620149045447e-05,
+      "loss": 3.4026,
+      "step": 5290
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.555778327631903e-05,
+      "loss": 3.5597,
+      "step": 5295
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.554935744964734e-05,
+      "loss": 3.4161,
+      "step": 5300
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.55409240118474e-05,
+      "loss": 3.5362,
+      "step": 5305
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.553248296432835e-05,
+      "loss": 3.4874,
+      "step": 5310
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.552403430850068e-05,
+      "loss": 3.5107,
+      "step": 5315
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.551557804577612e-05,
+      "loss": 3.4552,
+      "step": 5320
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.550711417756767e-05,
+      "loss": 3.389,
+      "step": 5325
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.549864270528965e-05,
+      "loss": 5.7899,
+      "step": 5330
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.549016363035758e-05,
+      "loss": 3.6934,
+      "step": 5335
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.548167695418828e-05,
+      "loss": 3.5228,
+      "step": 5340
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.547318267819986e-05,
+      "loss": 3.5978,
+      "step": 5345
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.546468080381167e-05,
+      "loss": 3.5975,
+      "step": 5350
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.545617133244436e-05,
+      "loss": 3.5186,
+      "step": 5355
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.544765426551981e-05,
+      "loss": 3.5706,
+      "step": 5360
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.543912960446117e-05,
+      "loss": 3.3869,
+      "step": 5365
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.543059735069293e-05,
+      "loss": 3.5115,
+      "step": 5370
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.542205750564077e-05,
+      "loss": 3.5062,
+      "step": 5375
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.541351007073167e-05,
+      "loss": 3.4654,
+      "step": 5380
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.540495504739387e-05,
+      "loss": 3.4599,
+      "step": 5385
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.53963924370569e-05,
+      "loss": 3.595,
+      "step": 5390
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.538782224115151e-05,
+      "loss": 3.6474,
+      "step": 5395
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.537924446110976e-05,
+      "loss": 3.5036,
+      "step": 5400
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.537065909836497e-05,
+      "loss": 3.4549,
+      "step": 5405
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.536206615435174e-05,
+      "loss": 3.5098,
+      "step": 5410
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.535346563050591e-05,
+      "loss": 3.5734,
+      "step": 5415
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.534485752826456e-05,
+      "loss": 3.4922,
+      "step": 5420
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.53362418490661e-05,
+      "loss": 3.5795,
+      "step": 5425
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.53276185943502e-05,
+      "loss": 3.5053,
+      "step": 5430
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.531898776555773e-05,
+      "loss": 3.4518,
+      "step": 5435
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.531034936413091e-05,
+      "loss": 3.472,
+      "step": 5440
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.530170339151314e-05,
+      "loss": 3.6654,
+      "step": 5445
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.529304984914917e-05,
+      "loss": 3.6008,
+      "step": 5450
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.528438873848496e-05,
+      "loss": 3.5164,
+      "step": 5455
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.527572006096773e-05,
+      "loss": 3.4701,
+      "step": 5460
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.526704381804601e-05,
+      "loss": 3.4744,
+      "step": 5465
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.525836001116957e-05,
+      "loss": 3.5249,
+      "step": 5470
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.524966864178943e-05,
+      "loss": 3.4724,
+      "step": 5475
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.524096971135789e-05,
+      "loss": 3.6023,
+      "step": 5480
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.52322632213285e-05,
+      "loss": 3.6178,
+      "step": 5485
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.52235491731561e-05,
+      "loss": 3.5179,
+      "step": 5490
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.521482756829675e-05,
+      "loss": 3.4906,
+      "step": 5495
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.520609840820782e-05,
+      "loss": 3.489,
+      "step": 5500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.51973616943479e-05,
+      "loss": 3.3993,
+      "step": 5505
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.518861742817689e-05,
+      "loss": 3.4716,
+      "step": 5510
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.51798656111559e-05,
+      "loss": 3.5395,
+      "step": 5515
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.517110624474733e-05,
+      "loss": 3.5444,
+      "step": 5520
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.516233933041485e-05,
+      "loss": 3.4727,
+      "step": 5525
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.515356486962337e-05,
+      "loss": 3.5875,
+      "step": 5530
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.514478286383907e-05,
+      "loss": 3.5164,
+      "step": 5535
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.513599331452939e-05,
+      "loss": 3.5834,
+      "step": 5540
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.512719622316304e-05,
+      "loss": 3.5212,
+      "step": 5545
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.511839159120997e-05,
+      "loss": 3.4734,
+      "step": 5550
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.51095794201414e-05,
+      "loss": 3.5318,
+      "step": 5555
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.510075971142983e-05,
+      "loss": 3.5362,
+      "step": 5560
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.509193246654896e-05,
+      "loss": 3.4287,
+      "step": 5565
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.508309768697384e-05,
+      "loss": 3.4798,
+      "step": 5570
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.507425537418068e-05,
+      "loss": 3.4357,
+      "step": 5575
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.506540552964704e-05,
+      "loss": 3.4741,
+      "step": 5580
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.505654815485168e-05,
+      "loss": 3.469,
+      "step": 5585
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.504768325127464e-05,
+      "loss": 3.5751,
+      "step": 5590
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.503881082039719e-05,
+      "loss": 3.4429,
+      "step": 5595
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.50299308637019e-05,
+      "loss": 3.4897,
+      "step": 5600
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.502104338267257e-05,
+      "loss": 3.5549,
+      "step": 5605
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.501214837879428e-05,
+      "loss": 3.5082,
+      "step": 5610
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.500324585355333e-05,
+      "loss": 3.5454,
+      "step": 5615
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.499433580843732e-05,
+      "loss": 3.5155,
+      "step": 5620
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.498541824493506e-05,
+      "loss": 3.5166,
+      "step": 5625
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.497649316453667e-05,
+      "loss": 3.4178,
+      "step": 5630
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.496756056873348e-05,
+      "loss": 3.4867,
+      "step": 5635
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.49586204590181e-05,
+      "loss": 3.4911,
+      "step": 5640
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.494967283688438e-05,
+      "loss": 3.4775,
+      "step": 5645
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.494071770382745e-05,
+      "loss": 3.533,
+      "step": 5650
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.493175506134367e-05,
+      "loss": 3.5321,
+      "step": 5655
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.492278491093067e-05,
+      "loss": 3.5855,
+      "step": 5660
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.491380725408731e-05,
+      "loss": 3.4522,
+      "step": 5665
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.490482209231374e-05,
+      "loss": 3.4491,
+      "step": 5670
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.489582942711135e-05,
+      "loss": 3.4605,
+      "step": 5675
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.488682925998277e-05,
+      "loss": 3.4448,
+      "step": 5680
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.48778215924319e-05,
+      "loss": 3.4705,
+      "step": 5685
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.48688064259639e-05,
+      "loss": 3.6538,
+      "step": 5690
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.485978376208513e-05,
+      "loss": 3.4914,
+      "step": 5695
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.485075360230328e-05,
+      "loss": 3.469,
+      "step": 5700
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.484171594812728e-05,
+      "loss": 3.4848,
+      "step": 5705
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.483267080106723e-05,
+      "loss": 3.5486,
+      "step": 5710
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.482361816263456e-05,
+      "loss": 3.4274,
+      "step": 5715
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.481455803434195e-05,
+      "loss": 3.4287,
+      "step": 5720
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.48054904177033e-05,
+      "loss": 3.4027,
+      "step": 5725
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.479641531423377e-05,
+      "loss": 3.476,
+      "step": 5730
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.47873327254498e-05,
+      "loss": 3.4313,
+      "step": 5735
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.477824265286904e-05,
+      "loss": 3.5499,
+      "step": 5740
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.476914509801039e-05,
+      "loss": 3.5445,
+      "step": 5745
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.476004006239406e-05,
+      "loss": 3.4624,
+      "step": 5750
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.475092754754142e-05,
+      "loss": 3.38,
+      "step": 5755
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.474180755497517e-05,
+      "loss": 3.4972,
+      "step": 5760
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.473268008621922e-05,
+      "loss": 3.5249,
+      "step": 5765
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.472354514279874e-05,
+      "loss": 3.3467,
+      "step": 5770
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.471440272624012e-05,
+      "loss": 3.3703,
+      "step": 5775
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.470525283807105e-05,
+      "loss": 3.5218,
+      "step": 5780
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.469609547982044e-05,
+      "loss": 3.4717,
+      "step": 5785
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.468693065301843e-05,
+      "loss": 3.5488,
+      "step": 5790
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.467775835919646e-05,
+      "loss": 3.5225,
+      "step": 5795
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.466857859988715e-05,
+      "loss": 3.3677,
+      "step": 5800
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.465939137662444e-05,
+      "loss": 3.4854,
+      "step": 5805
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.465019669094346e-05,
+      "loss": 3.4977,
+      "step": 5810
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.464099454438062e-05,
+      "loss": 3.5181,
+      "step": 5815
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.463178493847355e-05,
+      "loss": 3.51,
+      "step": 5820
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.462256787476113e-05,
+      "loss": 3.4597,
+      "step": 5825
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.461334335478354e-05,
+      "loss": 3.4741,
+      "step": 5830
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.460411138008214e-05,
+      "loss": 3.4633,
+      "step": 5835
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.459487195219954e-05,
+      "loss": 3.443,
+      "step": 5840
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.458562507267963e-05,
+      "loss": 3.532,
+      "step": 5845
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.457637074306754e-05,
+      "loss": 3.446,
+      "step": 5850
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.456710896490961e-05,
+      "loss": 3.4672,
+      "step": 5855
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.455783973975348e-05,
+      "loss": 3.5168,
+      "step": 5860
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.454856306914797e-05,
+      "loss": 3.4872,
+      "step": 5865
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.453927895464321e-05,
+      "loss": 3.4995,
+      "step": 5870
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.452998739779051e-05,
+      "loss": 3.5721,
+      "step": 5875
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.452068840014246e-05,
+      "loss": 3.4878,
+      "step": 5880
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.451138196325291e-05,
+      "loss": 3.6362,
+      "step": 5885
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.45020680886769e-05,
+      "loss": 3.6192,
+      "step": 5890
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.449274677797078e-05,
+      "loss": 3.5701,
+      "step": 5895
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.448341803269208e-05,
+      "loss": 3.4613,
+      "step": 5900
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.44740818543996e-05,
+      "loss": 3.5167,
+      "step": 5905
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.44647382446534e-05,
+      "loss": 3.4528,
+      "step": 5910
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.445538720501474e-05,
+      "loss": 3.4874,
+      "step": 5915
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.444602873704616e-05,
+      "loss": 3.3871,
+      "step": 5920
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.443666284231141e-05,
+      "loss": 3.358,
+      "step": 5925
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.44272895223755e-05,
+      "loss": 3.4682,
+      "step": 5930
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.441790877880469e-05,
+      "loss": 3.4824,
+      "step": 5935
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.440852061316645e-05,
+      "loss": 3.4733,
+      "step": 5940
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.439912502702952e-05,
+      "loss": 3.4569,
+      "step": 5945
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.438972202196387e-05,
+      "loss": 3.556,
+      "step": 5950
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.43803115995407e-05,
+      "loss": 3.4199,
+      "step": 5955
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.437089376133246e-05,
+      "loss": 3.4762,
+      "step": 5960
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.436146850891283e-05,
+      "loss": 3.4722,
+      "step": 5965
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.435203584385673e-05,
+      "loss": 3.4973,
+      "step": 5970
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.434259576774034e-05,
+      "loss": 3.4513,
+      "step": 5975
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.433314828214103e-05,
+      "loss": 3.446,
+      "step": 5980
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.43236933886375e-05,
+      "loss": 3.4277,
+      "step": 5985
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.431423108880957e-05,
+      "loss": 3.4547,
+      "step": 5990
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.430476138423838e-05,
+      "loss": 3.4671,
+      "step": 5995
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.429528427650627e-05,
+      "loss": 3.3838,
+      "step": 6000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.428579976719684e-05,
+      "loss": 3.537,
+      "step": 6005
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.427630785789491e-05,
+      "loss": 3.5579,
+      "step": 6010
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.426680855018654e-05,
+      "loss": 3.4802,
+      "step": 6015
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.425730184565907e-05,
+      "loss": 3.6951,
+      "step": 6020
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.424778774590096e-05,
+      "loss": 3.5948,
+      "step": 6025
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.423826625250204e-05,
+      "loss": 3.4322,
+      "step": 6030
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.422873736705329e-05,
+      "loss": 3.502,
+      "step": 6035
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.421920109114696e-05,
+      "loss": 3.4098,
+      "step": 6040
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.420965742637653e-05,
+      "loss": 3.3394,
+      "step": 6045
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.420010637433669e-05,
+      "loss": 3.4444,
+      "step": 6050
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.419054793662342e-05,
+      "loss": 3.4101,
+      "step": 6055
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.418098211483387e-05,
+      "loss": 3.3815,
+      "step": 6060
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.417140891056646e-05,
+      "loss": 3.4175,
+      "step": 6065
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.416182832542085e-05,
+      "loss": 3.3747,
+      "step": 6070
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.41522403609979e-05,
+      "loss": 3.4681,
+      "step": 6075
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.414264501889975e-05,
+      "loss": 3.431,
+      "step": 6080
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.41330423007297e-05,
+      "loss": 3.4742,
+      "step": 6085
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.412343220809239e-05,
+      "loss": 3.5554,
+      "step": 6090
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.41138147425936e-05,
+      "loss": 3.4382,
+      "step": 6095
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.410418990584034e-05,
+      "loss": 3.472,
+      "step": 6100
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.409455769944093e-05,
+      "loss": 3.3726,
+      "step": 6105
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.408491812500486e-05,
+      "loss": 3.538,
+      "step": 6110
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.407527118414287e-05,
+      "loss": 3.3996,
+      "step": 6115
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.406561687846693e-05,
+      "loss": 3.4461,
+      "step": 6120
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.405595520959025e-05,
+      "loss": 3.4838,
+      "step": 6125
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.404628617912723e-05,
+      "loss": 3.3522,
+      "step": 6130
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.403660978869353e-05,
+      "loss": 3.4166,
+      "step": 6135
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.402692603990608e-05,
+      "loss": 3.4069,
+      "step": 6140
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.401723493438295e-05,
+      "loss": 3.4151,
+      "step": 6145
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.400753647374352e-05,
+      "loss": 3.396,
+      "step": 6150
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.399783065960835e-05,
+      "loss": 3.3092,
+      "step": 6155
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.398811749359924e-05,
+      "loss": 3.5044,
+      "step": 6160
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.397839697733924e-05,
+      "loss": 3.5053,
+      "step": 6165
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.396866911245261e-05,
+      "loss": 3.3604,
+      "step": 6170
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.395893390056484e-05,
+      "loss": 3.514,
+      "step": 6175
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.394919134330264e-05,
+      "loss": 3.5694,
+      "step": 6180
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.393944144229396e-05,
+      "loss": 3.3524,
+      "step": 6185
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.392968419916797e-05,
+      "loss": 3.4525,
+      "step": 6190
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.391991961555507e-05,
+      "loss": 3.4743,
+      "step": 6195
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.39101476930869e-05,
+      "loss": 3.463,
+      "step": 6200
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.390036843339629e-05,
+      "loss": 3.4072,
+      "step": 6205
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.389058183811732e-05,
+      "loss": 3.4312,
+      "step": 6210
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.38807879088853e-05,
+      "loss": 3.4904,
+      "step": 6215
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.387098664733676e-05,
+      "loss": 3.4388,
+      "step": 6220
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.386117805510945e-05,
+      "loss": 3.4304,
+      "step": 6225
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.385136213384237e-05,
+      "loss": 3.3797,
+      "step": 6230
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.38415388851757e-05,
+      "loss": 3.5173,
+      "step": 6235
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.383170831075086e-05,
+      "loss": 3.5454,
+      "step": 6240
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.382187041221054e-05,
+      "loss": 3.4353,
+      "step": 6245
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.381202519119858e-05,
+      "loss": 3.2827,
+      "step": 6250
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.38021726493601e-05,
+      "loss": 3.479,
+      "step": 6255
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.37923127883414e-05,
+      "loss": 3.3865,
+      "step": 6260
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.378244560979007e-05,
+      "loss": 3.4672,
+      "step": 6265
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.377257111535482e-05,
+      "loss": 3.5214,
+      "step": 6270
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.376268930668569e-05,
+      "loss": 3.5744,
+      "step": 6275
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.375280018543387e-05,
+      "loss": 3.4361,
+      "step": 6280
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.374290375325181e-05,
+      "loss": 3.5391,
+      "step": 6285
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.373300001179316e-05,
+      "loss": 3.4294,
+      "step": 6290
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.37230889627128e-05,
+      "loss": 3.4034,
+      "step": 6295
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.371317060766682e-05,
+      "loss": 3.4662,
+      "step": 6300
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.370324494831254e-05,
+      "loss": 3.3901,
+      "step": 6305
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.369331198630852e-05,
+      "loss": 3.5181,
+      "step": 6310
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.36833717233145e-05,
+      "loss": 3.3999,
+      "step": 6315
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.367342416099147e-05,
+      "loss": 3.3726,
+      "step": 6320
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.366346930100163e-05,
+      "loss": 3.4302,
+      "step": 6325
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.36535071450084e-05,
+      "loss": 3.4255,
+      "step": 6330
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.364353769467643e-05,
+      "loss": 3.3785,
+      "step": 6335
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.363356095167156e-05,
+      "loss": 3.3677,
+      "step": 6340
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.362357691766088e-05,
+      "loss": 3.3898,
+      "step": 6345
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.361358559431266e-05,
+      "loss": 3.4866,
+      "step": 6350
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.360358698329646e-05,
+      "loss": 3.4365,
+      "step": 6355
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.359358108628298e-05,
+      "loss": 3.3719,
+      "step": 6360
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.358356790494416e-05,
+      "loss": 3.3976,
+      "step": 6365
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.357354744095317e-05,
+      "loss": 3.3277,
+      "step": 6370
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.35635196959844e-05,
+      "loss": 3.4963,
+      "step": 6375
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.355348467171347e-05,
+      "loss": 3.3767,
+      "step": 6380
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.354344236981716e-05,
+      "loss": 3.3707,
+      "step": 6385
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.353339279197351e-05,
+      "loss": 3.4419,
+      "step": 6390
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.352333593986176e-05,
+      "loss": 3.4277,
+      "step": 6395
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.35132718151624e-05,
+      "loss": 3.4488,
+      "step": 6400
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.35032004195571e-05,
+      "loss": 3.4538,
+      "step": 6405
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.349312175472873e-05,
+      "loss": 3.5245,
+      "step": 6410
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.348303582236142e-05,
+      "loss": 3.3881,
+      "step": 6415
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.347294262414047e-05,
+      "loss": 3.3269,
+      "step": 6420
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.346284216175245e-05,
+      "loss": 3.5113,
+      "step": 6425
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.345273443688509e-05,
+      "loss": 3.3841,
+      "step": 6430
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.344261945122734e-05,
+      "loss": 3.3032,
+      "step": 6435
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.343249720646939e-05,
+      "loss": 3.3795,
+      "step": 6440
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.342236770430264e-05,
+      "loss": 3.4693,
+      "step": 6445
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.341223094641967e-05,
+      "loss": 3.3738,
+      "step": 6450
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.340208693451431e-05,
+      "loss": 3.5182,
+      "step": 6455
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.339193567028157e-05,
+      "loss": 3.3895,
+      "step": 6460
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.338177715541772e-05,
+      "loss": 3.4105,
+      "step": 6465
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.337161139162019e-05,
+      "loss": 3.4522,
+      "step": 6470
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.336143838058761e-05,
+      "loss": 3.4275,
+      "step": 6475
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.335125812401993e-05,
+      "loss": 3.4205,
+      "step": 6480
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.334107062361816e-05,
+      "loss": 3.3925,
+      "step": 6485
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.333087588108462e-05,
+      "loss": 3.3985,
+      "step": 6490
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.332067389812282e-05,
+      "loss": 3.5249,
+      "step": 6495
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.331046467643746e-05,
+      "loss": 3.4853,
+      "step": 6500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.330024821773448e-05,
+      "loss": 3.3939,
+      "step": 6505
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.329002452372102e-05,
+      "loss": 3.4135,
+      "step": 6510
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.327979359610537e-05,
+      "loss": 3.4721,
+      "step": 6515
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.326955543659714e-05,
+      "loss": 3.3838,
+      "step": 6520
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.325931004690706e-05,
+      "loss": 3.2852,
+      "step": 6525
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.324905742874709e-05,
+      "loss": 3.4143,
+      "step": 6530
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.323879758383042e-05,
+      "loss": 3.3898,
+      "step": 6535
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.322853051387141e-05,
+      "loss": 3.3595,
+      "step": 6540
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.321825622058568e-05,
+      "loss": 3.471,
+      "step": 6545
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.320797470569001e-05,
+      "loss": 3.4023,
+      "step": 6550
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.319768597090242e-05,
+      "loss": 3.3702,
+      "step": 6555
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.318739001794208e-05,
+      "loss": 3.4115,
+      "step": 6560
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.317708684852942e-05,
+      "loss": 3.4272,
+      "step": 6565
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.31667764643861e-05,
+      "loss": 3.3508,
+      "step": 6570
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.31564588672349e-05,
+      "loss": 3.3994,
+      "step": 6575
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.314613405879986e-05,
+      "loss": 3.3722,
+      "step": 6580
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.313580204080623e-05,
+      "loss": 3.3133,
+      "step": 6585
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.312546281498047e-05,
+      "loss": 3.5068,
+      "step": 6590
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.311511638305018e-05,
+      "loss": 3.6434,
+      "step": 6595
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.310476274674425e-05,
+      "loss": 3.3913,
+      "step": 6600
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.30944019077927e-05,
+      "loss": 3.2909,
+      "step": 6605
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.30840338679268e-05,
+      "loss": 3.4189,
+      "step": 6610
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.307365862887905e-05,
+      "loss": 3.3926,
+      "step": 6615
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.306327619238306e-05,
+      "loss": 3.405,
+      "step": 6620
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.305288656017372e-05,
+      "loss": 3.4235,
+      "step": 6625
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.304248973398711e-05,
+      "loss": 3.3262,
+      "step": 6630
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.303208571556048e-05,
+      "loss": 3.3673,
+      "step": 6635
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.30216745066323e-05,
+      "loss": 3.5034,
+      "step": 6640
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.301125610894226e-05,
+      "loss": 3.4158,
+      "step": 6645
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.300083052423124e-05,
+      "loss": 3.3841,
+      "step": 6650
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.29903977542413e-05,
+      "loss": 3.6112,
+      "step": 6655
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.297995780071571e-05,
+      "loss": 3.4271,
+      "step": 6660
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.296951066539899e-05,
+      "loss": 3.4073,
+      "step": 6665
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.295905635003677e-05,
+      "loss": 3.4171,
+      "step": 6670
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.294859485637595e-05,
+      "loss": 3.4383,
+      "step": 6675
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.29381261861646e-05,
+      "loss": 3.4137,
+      "step": 6680
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.292765034115198e-05,
+      "loss": 3.4224,
+      "step": 6685
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.291716732308862e-05,
+      "loss": 3.4049,
+      "step": 6690
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.290667713372613e-05,
+      "loss": 3.3064,
+      "step": 6695
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.289617977481741e-05,
+      "loss": 3.5093,
+      "step": 6700
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.288567524811654e-05,
+      "loss": 3.3389,
+      "step": 6705
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.287516355537876e-05,
+      "loss": 3.2882,
+      "step": 6710
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.286464469836056e-05,
+      "loss": 3.2947,
+      "step": 6715
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.28541186788196e-05,
+      "loss": 3.327,
+      "step": 6720
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.284358549851471e-05,
+      "loss": 3.4484,
+      "step": 6725
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.283304515920598e-05,
+      "loss": 3.4053,
+      "step": 6730
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.282249766265464e-05,
+      "loss": 3.3557,
+      "step": 6735
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.281194301062315e-05,
+      "loss": 3.4444,
+      "step": 6740
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.280138120487514e-05,
+      "loss": 3.4305,
+      "step": 6745
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.279081224717547e-05,
+      "loss": 3.4086,
+      "step": 6750
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.278023613929014e-05,
+      "loss": 3.3945,
+      "step": 6755
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.276965288298642e-05,
+      "loss": 3.4716,
+      "step": 6760
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.27590624800327e-05,
+      "loss": 3.3553,
+      "step": 6765
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.27484649321986e-05,
+      "loss": 3.3672,
+      "step": 6770
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.273786024125496e-05,
+      "loss": 3.4828,
+      "step": 6775
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.272724840897375e-05,
+      "loss": 3.4378,
+      "step": 6780
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.271662943712818e-05,
+      "loss": 3.368,
+      "step": 6785
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.270600332749264e-05,
+      "loss": 3.4185,
+      "step": 6790
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.269537008184273e-05,
+      "loss": 3.4345,
+      "step": 6795
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.268472970195519e-05,
+      "loss": 3.4419,
+      "step": 6800
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.267408218960803e-05,
+      "loss": 3.4055,
+      "step": 6805
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.266342754658036e-05,
+      "loss": 3.3949,
+      "step": 6810
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.265276577465259e-05,
+      "loss": 3.2855,
+      "step": 6815
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.264209687560621e-05,
+      "loss": 3.458,
+      "step": 6820
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.263142085122399e-05,
+      "loss": 3.357,
+      "step": 6825
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.262073770328983e-05,
+      "loss": 3.3464,
+      "step": 6830
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.261004743358886e-05,
+      "loss": 3.303,
+      "step": 6835
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.259935004390737e-05,
+      "loss": 3.3998,
+      "step": 6840
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.258864553603286e-05,
+      "loss": 3.4927,
+      "step": 6845
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.257793391175402e-05,
+      "loss": 3.4475,
+      "step": 6850
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.256721517286073e-05,
+      "loss": 3.4236,
+      "step": 6855
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.255648932114403e-05,
+      "loss": 3.4478,
+      "step": 6860
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.254575635839618e-05,
+      "loss": 3.3146,
+      "step": 6865
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.253501628641061e-05,
+      "loss": 3.3156,
+      "step": 6870
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.252426910698198e-05,
+      "loss": 3.51,
+      "step": 6875
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.251351482190604e-05,
+      "loss": 3.4224,
+      "step": 6880
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.250275343297985e-05,
+      "loss": 3.469,
+      "step": 6885
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.249198494200158e-05,
+      "loss": 3.44,
+      "step": 6890
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.24812093507706e-05,
+      "loss": 3.4528,
+      "step": 6895
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.247042666108747e-05,
+      "loss": 3.397,
+      "step": 6900
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.245963687475395e-05,
+      "loss": 3.3916,
+      "step": 6905
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.244883999357293e-05,
+      "loss": 3.3421,
+      "step": 6910
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.24380360193486e-05,
+      "loss": 3.304,
+      "step": 6915
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.242722495388621e-05,
+      "loss": 3.4458,
+      "step": 6920
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.241640679899226e-05,
+      "loss": 3.4959,
+      "step": 6925
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.240558155647443e-05,
+      "loss": 3.3509,
+      "step": 6930
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.239474922814157e-05,
+      "loss": 3.4062,
+      "step": 6935
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.238390981580375e-05,
+      "loss": 3.3638,
+      "step": 6940
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.237306332127215e-05,
+      "loss": 3.5189,
+      "step": 6945
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.236220974635923e-05,
+      "loss": 3.4664,
+      "step": 6950
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.235134909287853e-05,
+      "loss": 3.3441,
+      "step": 6955
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.234048136264485e-05,
+      "loss": 3.3271,
+      "step": 6960
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.232960655747416e-05,
+      "loss": 3.3432,
+      "step": 6965
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.231872467918358e-05,
+      "loss": 3.3183,
+      "step": 6970
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.230783572959143e-05,
+      "loss": 3.4438,
+      "step": 6975
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.229693971051723e-05,
+      "loss": 3.4864,
+      "step": 6980
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.228603662378166e-05,
+      "loss": 3.3551,
+      "step": 6985
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.227512647120656e-05,
+      "loss": 3.3426,
+      "step": 6990
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.2264209254615e-05,
+      "loss": 3.3719,
+      "step": 6995
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.225328497583122e-05,
+      "loss": 3.431,
+      "step": 7000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.224235363668058e-05,
+      "loss": 3.338,
+      "step": 7005
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.223141523898972e-05,
+      "loss": 3.4871,
+      "step": 7010
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.222046978458639e-05,
+      "loss": 3.409,
+      "step": 7015
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.22095172752995e-05,
+      "loss": 3.3695,
+      "step": 7020
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.21985577129592e-05,
+      "loss": 3.2672,
+      "step": 7025
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.218759109939681e-05,
+      "loss": 3.2872,
+      "step": 7030
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.217661743644477e-05,
+      "loss": 3.3244,
+      "step": 7035
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.216563672593677e-05,
+      "loss": 3.296,
+      "step": 7040
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.215464896970763e-05,
+      "loss": 3.3604,
+      "step": 7045
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.214365416959336e-05,
+      "loss": 3.537,
+      "step": 7050
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.213265232743119e-05,
+      "loss": 3.3155,
+      "step": 7055
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.212164344505942e-05,
+      "loss": 3.3868,
+      "step": 7060
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.211062752431765e-05,
+      "loss": 3.3851,
+      "step": 7065
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.209960456704656e-05,
+      "loss": 3.3645,
+      "step": 7070
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.208857457508809e-05,
+      "loss": 3.3106,
+      "step": 7075
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.207753755028526e-05,
+      "loss": 3.455,
+      "step": 7080
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.206649349448235e-05,
+      "loss": 3.4107,
+      "step": 7085
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.205544240952477e-05,
+      "loss": 3.3842,
+      "step": 7090
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.204438429725911e-05,
+      "loss": 3.3339,
+      "step": 7095
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.203331915953314e-05,
+      "loss": 3.4014,
+      "step": 7100
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.202224699819584e-05,
+      "loss": 3.2919,
+      "step": 7105
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.201116781509726e-05,
+      "loss": 3.4239,
+      "step": 7110
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.200008161208875e-05,
+      "loss": 3.3587,
+      "step": 7115
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.198898839102273e-05,
+      "loss": 3.4071,
+      "step": 7120
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.197788815375287e-05,
+      "loss": 3.3098,
+      "step": 7125
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.196678090213395e-05,
+      "loss": 3.4032,
+      "step": 7130
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.195566663802198e-05,
+      "loss": 3.3555,
+      "step": 7135
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.194454536327408e-05,
+      "loss": 3.3467,
+      "step": 7140
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.19334170797486e-05,
+      "loss": 3.5143,
+      "step": 7145
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.192228178930503e-05,
+      "loss": 3.5187,
+      "step": 7150
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.191113949380402e-05,
+      "loss": 3.3433,
+      "step": 7155
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.189999019510742e-05,
+      "loss": 3.3817,
+      "step": 7160
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.188883389507823e-05,
+      "loss": 3.3413,
+      "step": 7165
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.187767059558064e-05,
+      "loss": 3.4649,
+      "step": 7170
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.186650029847998e-05,
+      "loss": 3.4007,
+      "step": 7175
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.185532300564276e-05,
+      "loss": 3.3584,
+      "step": 7180
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.184413871893668e-05,
+      "loss": 3.3196,
+      "step": 7185
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.183294744023058e-05,
+      "loss": 3.3407,
+      "step": 7190
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.18217491713945e-05,
+      "loss": 3.4793,
+      "step": 7195
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.181054391429962e-05,
+      "loss": 3.3154,
+      "step": 7200
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.179933167081828e-05,
+      "loss": 3.4615,
+      "step": 7205
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.178811244282404e-05,
+      "loss": 3.303,
+      "step": 7210
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.177688623219154e-05,
+      "loss": 3.4037,
+      "step": 7215
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.176565304079669e-05,
+      "loss": 3.2974,
+      "step": 7220
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.175441287051648e-05,
+      "loss": 3.3506,
+      "step": 7225
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.174316572322912e-05,
+      "loss": 3.3627,
+      "step": 7230
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.173191160081395e-05,
+      "loss": 3.3045,
+      "step": 7235
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.17206505051515e-05,
+      "loss": 3.4454,
+      "step": 7240
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.170938243812349e-05,
+      "loss": 3.3738,
+      "step": 7245
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.16981074016127e-05,
+      "loss": 3.3887,
+      "step": 7250
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.16868253975032e-05,
+      "loss": 3.3158,
+      "step": 7255
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.167553642768017e-05,
+      "loss": 3.3204,
+      "step": 7260
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.166424049402993e-05,
+      "loss": 3.2622,
+      "step": 7265
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.165293759844e-05,
+      "loss": 3.4392,
+      "step": 7270
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.164162774279905e-05,
+      "loss": 3.3696,
+      "step": 7275
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.163031092899693e-05,
+      "loss": 3.32,
+      "step": 7280
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.161898715892463e-05,
+      "loss": 3.413,
+      "step": 7285
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.160765643447429e-05,
+      "loss": 3.4235,
+      "step": 7290
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.159631875753925e-05,
+      "loss": 3.3994,
+      "step": 7295
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.1584974130014e-05,
+      "loss": 3.3566,
+      "step": 7300
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.157362255379419e-05,
+      "loss": 3.3768,
+      "step": 7305
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.156226403077657e-05,
+      "loss": 3.4625,
+      "step": 7310
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.155089856285917e-05,
+      "loss": 3.3889,
+      "step": 7315
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.153952615194111e-05,
+      "loss": 3.3396,
+      "step": 7320
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.152814679992265e-05,
+      "loss": 3.2677,
+      "step": 7325
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.151676050870526e-05,
+      "loss": 3.4753,
+      "step": 7330
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.150536728019154e-05,
+      "loss": 3.4862,
+      "step": 7335
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.149396711628523e-05,
+      "loss": 3.42,
+      "step": 7340
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.14825600188913e-05,
+      "loss": 3.3956,
+      "step": 7345
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.147114598991583e-05,
+      "loss": 3.3298,
+      "step": 7350
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.145972503126603e-05,
+      "loss": 3.4637,
+      "step": 7355
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.14482971448503e-05,
+      "loss": 3.3217,
+      "step": 7360
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.143686233257824e-05,
+      "loss": 3.39,
+      "step": 7365
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.142542059636053e-05,
+      "loss": 3.2514,
+      "step": 7370
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.141397193810903e-05,
+      "loss": 3.3802,
+      "step": 7375
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.14025163597368e-05,
+      "loss": 3.3096,
+      "step": 7380
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.1391053863158e-05,
+      "loss": 3.3008,
+      "step": 7385
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.137958445028799e-05,
+      "loss": 3.3681,
+      "step": 7390
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.136810812304326e-05,
+      "loss": 3.2792,
+      "step": 7395
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.135662488334147e-05,
+      "loss": 3.3396,
+      "step": 7400
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.13451347331014e-05,
+      "loss": 3.1766,
+      "step": 7405
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.133363767424303e-05,
+      "loss": 3.2791,
+      "step": 7410
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.132213370868748e-05,
+      "loss": 3.2549,
+      "step": 7415
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.131062283835701e-05,
+      "loss": 3.3125,
+      "step": 7420
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.129910506517503e-05,
+      "loss": 3.418,
+      "step": 7425
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.128758039106616e-05,
+      "loss": 3.4386,
+      "step": 7430
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.127604881795609e-05,
+      "loss": 3.4677,
+      "step": 7435
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.126451034777173e-05,
+      "loss": 3.298,
+      "step": 7440
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.12529649824411e-05,
+      "loss": 3.4341,
+      "step": 7445
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.124141272389337e-05,
+      "loss": 3.423,
+      "step": 7450
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.122985357405892e-05,
+      "loss": 3.3271,
+      "step": 7455
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.121828753486921e-05,
+      "loss": 3.2889,
+      "step": 7460
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.120671460825691e-05,
+      "loss": 3.2936,
+      "step": 7465
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.119513479615577e-05,
+      "loss": 3.3232,
+      "step": 7470
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.118354810050079e-05,
+      "loss": 3.3352,
+      "step": 7475
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.117195452322801e-05,
+      "loss": 3.3721,
+      "step": 7480
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.11603540662747e-05,
+      "loss": 3.3693,
+      "step": 7485
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.114874673157927e-05,
+      "loss": 3.3165,
+      "step": 7490
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.113713252108123e-05,
+      "loss": 3.3268,
+      "step": 7495
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.112551143672127e-05,
+      "loss": 3.3677,
+      "step": 7500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.111388348044125e-05,
+      "loss": 3.3763,
+      "step": 7505
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.110224865418417e-05,
+      "loss": 3.445,
+      "step": 7510
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.109060695989413e-05,
+      "loss": 3.3564,
+      "step": 7515
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.107895839951643e-05,
+      "loss": 3.4092,
+      "step": 7520
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.106730297499752e-05,
+      "loss": 3.3943,
+      "step": 7525
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.105564068828497e-05,
+      "loss": 3.2852,
+      "step": 7530
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.104397154132747e-05,
+      "loss": 3.3353,
+      "step": 7535
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.103229553607493e-05,
+      "loss": 3.2967,
+      "step": 7540
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.102061267447835e-05,
+      "loss": 3.3766,
+      "step": 7545
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.100892295848989e-05,
+      "loss": 3.3757,
+      "step": 7550
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.099722639006287e-05,
+      "loss": 3.3667,
+      "step": 7555
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.098552297115175e-05,
+      "loss": 3.2926,
+      "step": 7560
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.097381270371212e-05,
+      "loss": 3.328,
+      "step": 7565
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.09620955897007e-05,
+      "loss": 3.2686,
+      "step": 7570
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.09503716310754e-05,
+      "loss": 3.4043,
+      "step": 7575
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.093864082979525e-05,
+      "loss": 3.2797,
+      "step": 7580
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.092690318782042e-05,
+      "loss": 3.3132,
+      "step": 7585
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.091515870711221e-05,
+      "loss": 3.3585,
+      "step": 7590
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.09034073896331e-05,
+      "loss": 3.3959,
+      "step": 7595
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.08916492373467e-05,
+      "loss": 3.3246,
+      "step": 7600
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.087988425221772e-05,
+      "loss": 3.289,
+      "step": 7605
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.086811243621206e-05,
+      "loss": 3.313,
+      "step": 7610
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.085633379129677e-05,
+      "loss": 3.3358,
+      "step": 7615
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.084454831943998e-05,
+      "loss": 3.27,
+      "step": 7620
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.083275602261101e-05,
+      "loss": 3.3503,
+      "step": 7625
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.082095690278034e-05,
+      "loss": 3.2884,
+      "step": 7630
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.080915096191951e-05,
+      "loss": 3.3839,
+      "step": 7635
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.07973382020013e-05,
+      "loss": 3.3918,
+      "step": 7640
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.078551862499953e-05,
+      "loss": 3.338,
+      "step": 7645
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.077369223288923e-05,
+      "loss": 3.3628,
+      "step": 7650
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.076185902764656e-05,
+      "loss": 3.3756,
+      "step": 7655
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.075001901124878e-05,
+      "loss": 3.2336,
+      "step": 7660
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.073817218567431e-05,
+      "loss": 3.4262,
+      "step": 7665
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.072631855290275e-05,
+      "loss": 3.4738,
+      "step": 7670
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.071445811491475e-05,
+      "loss": 3.2575,
+      "step": 7675
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.070259087369218e-05,
+      "loss": 3.2508,
+      "step": 7680
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.069071683121799e-05,
+      "loss": 3.3256,
+      "step": 7685
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.067883598947628e-05,
+      "loss": 3.3984,
+      "step": 7690
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.066694835045233e-05,
+      "loss": 3.4846,
+      "step": 7695
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.06550539161325e-05,
+      "loss": 3.4769,
+      "step": 7700
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.06431526885043e-05,
+      "loss": 3.2745,
+      "step": 7705
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.063124466955639e-05,
+      "loss": 3.4091,
+      "step": 7710
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.061932986127857e-05,
+      "loss": 3.381,
+      "step": 7715
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.060740826566172e-05,
+      "loss": 3.4045,
+      "step": 7720
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.059547988469792e-05,
+      "loss": 3.3814,
+      "step": 7725
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.058354472038037e-05,
+      "loss": 3.2741,
+      "step": 7730
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.057160277470337e-05,
+      "loss": 3.2974,
+      "step": 7735
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.055965404966238e-05,
+      "loss": 3.3208,
+      "step": 7740
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.054769854725399e-05,
+      "loss": 3.3006,
+      "step": 7745
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.053573626947593e-05,
+      "loss": 3.3069,
+      "step": 7750
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.052376721832702e-05,
+      "loss": 3.3201,
+      "step": 7755
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.051179139580727e-05,
+      "loss": 3.3355,
+      "step": 7760
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.049980880391782e-05,
+      "loss": 3.2773,
+      "step": 7765
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.048781944466087e-05,
+      "loss": 3.3635,
+      "step": 7770
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.04758233200398e-05,
+      "loss": 3.4161,
+      "step": 7775
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.046382043205914e-05,
+      "loss": 3.3235,
+      "step": 7780
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.045181078272453e-05,
+      "loss": 3.4161,
+      "step": 7785
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.043979437404273e-05,
+      "loss": 3.36,
+      "step": 7790
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.042777120802164e-05,
+      "loss": 3.3585,
+      "step": 7795
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.041574128667027e-05,
+      "loss": 3.3419,
+      "step": 7800
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.040370461199879e-05,
+      "loss": 3.4033,
+      "step": 7805
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.039166118601849e-05,
+      "loss": 3.3773,
+      "step": 7810
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.037961101074175e-05,
+      "loss": 3.267,
+      "step": 7815
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.036755408818217e-05,
+      "loss": 3.2373,
+      "step": 7820
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.035549042035434e-05,
+      "loss": 3.319,
+      "step": 7825
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.03434200092741e-05,
+      "loss": 3.3202,
+      "step": 7830
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.033134285695837e-05,
+      "loss": 3.4028,
+      "step": 7835
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.031925896542521e-05,
+      "loss": 3.2446,
+      "step": 7840
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.030716833669375e-05,
+      "loss": 3.2822,
+      "step": 7845
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.029507097278433e-05,
+      "loss": 3.3783,
+      "step": 7850
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.028296687571834e-05,
+      "loss": 3.3665,
+      "step": 7855
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.027085604751835e-05,
+      "loss": 3.4438,
+      "step": 7860
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.025873849020805e-05,
+      "loss": 3.4149,
+      "step": 7865
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.024661420581222e-05,
+      "loss": 3.3443,
+      "step": 7870
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.023448319635676e-05,
+      "loss": 3.2593,
+      "step": 7875
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.022234546386877e-05,
+      "loss": 3.2595,
+      "step": 7880
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.021020101037637e-05,
+      "loss": 3.4701,
+      "step": 7885
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.019804983790888e-05,
+      "loss": 3.3207,
+      "step": 7890
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.018589194849672e-05,
+      "loss": 3.3839,
+      "step": 7895
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.01737273441714e-05,
+      "loss": 3.3502,
+      "step": 7900
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.016155602696559e-05,
+      "loss": 3.2952,
+      "step": 7905
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.01493779989131e-05,
+      "loss": 3.4568,
+      "step": 7910
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.013719326204881e-05,
+      "loss": 3.382,
+      "step": 7915
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.012500181840874e-05,
+      "loss": 3.3823,
+      "step": 7920
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.011280367003003e-05,
+      "loss": 3.2305,
+      "step": 7925
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.010059881895099e-05,
+      "loss": 3.2767,
+      "step": 7930
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.008838726721094e-05,
+      "loss": 3.3587,
+      "step": 7935
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.007616901685042e-05,
+      "loss": 3.3422,
+      "step": 7940
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.006394406991108e-05,
+      "loss": 3.2247,
+      "step": 7945
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.00517124284356e-05,
+      "loss": 3.2382,
+      "step": 7950
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.003947409446789e-05,
+      "loss": 3.3711,
+      "step": 7955
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.002722907005292e-05,
+      "loss": 3.4682,
+      "step": 7960
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.001497735723679e-05,
+      "loss": 3.3085,
+      "step": 7965
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.000271895806669e-05,
+      "loss": 3.382,
+      "step": 7970
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.9990453874591e-05,
+      "loss": 3.2384,
+      "step": 7975
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.997818210885912e-05,
+      "loss": 3.2619,
+      "step": 7980
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.996590366292164e-05,
+      "loss": 3.2424,
+      "step": 7985
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.995361853883026e-05,
+      "loss": 3.2358,
+      "step": 7990
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.994132673863774e-05,
+      "loss": 3.3097,
+      "step": 7995
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.992902826439802e-05,
+      "loss": 3.1725,
+      "step": 8000
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 38527,
+  "num_train_epochs": 1,
+  "save_steps": 1000,
+  "total_flos": 1.946643246297907e+19,
+  "trial_name": null,
+  "trial_params": null
+}