diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,7219 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.1557339285830159,
+  "eval_steps": 500,
+  "global_step": 6000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 5e-06,
+      "loss": 13.2358,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1e-05,
+      "loss": 12.2186,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5e-05,
+      "loss": 10.4519,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2e-05,
+      "loss": 9.1261,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5e-05,
+      "loss": 8.279,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3e-05,
+      "loss": 7.8981,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.5e-05,
+      "loss": 7.6883,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4e-05,
+      "loss": 7.4628,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.5e-05,
+      "loss": 7.17,
+      "step": 45
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5e-05,
+      "loss": 6.7632,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.500000000000001e-05,
+      "loss": 6.4632,
+      "step": 55
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6e-05,
+      "loss": 6.1266,
+      "step": 60
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.500000000000001e-05,
+      "loss": 5.7336,
+      "step": 65
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7e-05,
+      "loss": 5.4787,
+      "step": 70
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 5.2614,
+      "step": 75
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8e-05,
+      "loss": 5.0757,
+      "step": 80
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.5e-05,
+      "loss": 4.9078,
+      "step": 85
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9e-05,
+      "loss": 4.7294,
+      "step": 90
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.5e-05,
+      "loss": 4.5506,
+      "step": 95
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001,
+      "loss": 4.4312,
+      "step": 100
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999999582259295e-05,
+      "loss": 4.3942,
+      "step": 105
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999998329037248e-05,
+      "loss": 4.2683,
+      "step": 110
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999996240334068e-05,
+      "loss": 4.17,
+      "step": 115
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999993316150106e-05,
+      "loss": 4.1729,
+      "step": 120
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999989556485852e-05,
+      "loss": 4.0547,
+      "step": 125
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999984961341928e-05,
+      "loss": 3.9966,
+      "step": 130
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999979530719108e-05,
+      "loss": 3.9086,
+      "step": 135
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999973264618297e-05,
+      "loss": 3.8995,
+      "step": 140
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999966163040541e-05,
+      "loss": 3.7602,
+      "step": 145
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.99995822598703e-05,
+      "loss": 3.8154,
+      "step": 150
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999949453459088e-05,
+      "loss": 3.7233,
+      "step": 155
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999939845458179e-05,
+      "loss": 3.634,
+      "step": 160
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999929401985911e-05,
+      "loss": 3.7648,
+      "step": 165
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.99991812304403e-05,
+      "loss": 3.693,
+      "step": 170
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999906008634416e-05,
+      "loss": 3.6077,
+      "step": 175
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.9998930587591e-05,
+      "loss": 3.5466,
+      "step": 180
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.99987927342024e-05,
+      "loss": 3.6108,
+      "step": 185
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999864652620143e-05,
+      "loss": 3.5773,
+      "step": 190
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999849196361251e-05,
+      "loss": 3.4777,
+      "step": 195
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999832904646147e-05,
+      "loss": 3.337,
+      "step": 200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99981577747755e-05,
+      "loss": 3.5813,
+      "step": 205
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999797814858327e-05,
+      "loss": 3.4625,
+      "step": 210
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999779016791477e-05,
+      "loss": 3.4593,
+      "step": 215
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999759383280141e-05,
+      "loss": 3.4706,
+      "step": 220
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.9997389143276e-05,
+      "loss": 3.4355,
+      "step": 225
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999717609937275e-05,
+      "loss": 3.415,
+      "step": 230
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999695470112725e-05,
+      "loss": 3.4319,
+      "step": 235
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99967249485765e-05,
+      "loss": 3.3548,
+      "step": 240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999648684175888e-05,
+      "loss": 3.402,
+      "step": 245
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999624038071418e-05,
+      "loss": 3.3934,
+      "step": 250
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99959855654836e-05,
+      "loss": 3.3291,
+      "step": 255
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999572239610968e-05,
+      "loss": 3.2892,
+      "step": 260
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999545087263645e-05,
+      "loss": 3.4023,
+      "step": 265
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999517099510923e-05,
+      "loss": 3.3682,
+      "step": 270
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999488276357481e-05,
+      "loss": 3.3774,
+      "step": 275
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999458617808137e-05,
+      "loss": 3.3846,
+      "step": 280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999428123867843e-05,
+      "loss": 3.3378,
+      "step": 285
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999396794541695e-05,
+      "loss": 3.3754,
+      "step": 290
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999364629834931e-05,
+      "loss": 3.2837,
+      "step": 295
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999331629752924e-05,
+      "loss": 3.2658,
+      "step": 300
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999297794301188e-05,
+      "loss": 3.3225,
+      "step": 305
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999263123485376e-05,
+      "loss": 3.2257,
+      "step": 310
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999227617311282e-05,
+      "loss": 3.3045,
+      "step": 315
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99919127578484e-05,
+      "loss": 3.306,
+      "step": 320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999154098912122e-05,
+      "loss": 3.1919,
+      "step": 325
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999116086699338e-05,
+      "loss": 3.2254,
+      "step": 330
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999077239152843e-05,
+      "loss": 3.2447,
+      "step": 335
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999037556279127e-05,
+      "loss": 3.1712,
+      "step": 340
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998997038084818e-05,
+      "loss": 3.1948,
+      "step": 345
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998955684576692e-05,
+      "loss": 3.1788,
+      "step": 350
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998913495761654e-05,
+      "loss": 3.2006,
+      "step": 355
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998870471646757e-05,
+      "loss": 3.2676,
+      "step": 360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99882661223919e-05,
+      "loss": 3.1004,
+      "step": 365
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998781917546278e-05,
+      "loss": 3.1891,
+      "step": 370
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998736387575493e-05,
+      "loss": 3.2295,
+      "step": 375
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998690022334442e-05,
+      "loss": 3.1512,
+      "step": 380
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998642821830873e-05,
+      "loss": 3.1573,
+      "step": 385
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99859478607267e-05,
+      "loss": 3.1258,
+      "step": 390
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998545915067864e-05,
+      "loss": 3.0698,
+      "step": 395
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998496208824618e-05,
+      "loss": 3.2411,
+      "step": 400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99844566735124e-05,
+      "loss": 3.1722,
+      "step": 405
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998394290656172e-05,
+      "loss": 3.0864,
+      "step": 410
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998342078748002e-05,
+      "loss": 3.1235,
+      "step": 415
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998289031635453e-05,
+      "loss": 3.0942,
+      "step": 420
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998235149327389e-05,
+      "loss": 3.1132,
+      "step": 425
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998180431832816e-05,
+      "loss": 3.0869,
+      "step": 430
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998124879160872e-05,
+      "loss": 3.0515,
+      "step": 435
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998068491320844e-05,
+      "loss": 3.1459,
+      "step": 440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998011268322152e-05,
+      "loss": 3.025,
+      "step": 445
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997953210174361e-05,
+      "loss": 3.1462,
+      "step": 450
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997894316887167e-05,
+      "loss": 3.0696,
+      "step": 455
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997834588470414e-05,
+      "loss": 3.1428,
+      "step": 460
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997774024934082e-05,
+      "loss": 3.1769,
+      "step": 465
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997712626288292e-05,
+      "loss": 3.0412,
+      "step": 470
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997650392543303e-05,
+      "loss": 3.0462,
+      "step": 475
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997587323709512e-05,
+      "loss": 3.1061,
+      "step": 480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997523419797459e-05,
+      "loss": 3.0151,
+      "step": 485
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997458680817824e-05,
+      "loss": 3.0544,
+      "step": 490
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997393106781421e-05,
+      "loss": 3.041,
+      "step": 495
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99732669769921e-05,
+      "loss": 3.0779,
+      "step": 500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997259453582286e-05,
+      "loss": 3.0553,
+      "step": 505
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997191374441887e-05,
+      "loss": 2.9722,
+      "step": 510
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997122460289387e-05,
+      "loss": 2.9493,
+      "step": 515
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997052711136301e-05,
+      "loss": 3.1075,
+      "step": 520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996982126994287e-05,
+      "loss": 2.9499,
+      "step": 525
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996910707875136e-05,
+      "loss": 3.0209,
+      "step": 530
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996838453790784e-05,
+      "loss": 3.0253,
+      "step": 535
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996765364753302e-05,
+      "loss": 3.0425,
+      "step": 540
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996691440774906e-05,
+      "loss": 3.0457,
+      "step": 545
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996616681867945e-05,
+      "loss": 3.0292,
+      "step": 550
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996541088044913e-05,
+      "loss": 2.9844,
+      "step": 555
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996464659318442e-05,
+      "loss": 3.0697,
+      "step": 560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996387395701302e-05,
+      "loss": 2.9338,
+      "step": 565
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996309297206403e-05,
+      "loss": 2.9907,
+      "step": 570
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996230363846797e-05,
+      "loss": 3.0327,
+      "step": 575
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.996150595635671e-05,
+      "loss": 2.9725,
+      "step": 580
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.996069992586355e-05,
+      "loss": 3.0623,
+      "step": 585
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.995988554712318e-05,
+      "loss": 2.943,
+      "step": 590
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.995906282027166e-05,
+      "loss": 3.0203,
+      "step": 595
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.995823174544649e-05,
+      "loss": 3.0229,
+      "step": 600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.995739232278654e-05,
+      "loss": 2.9496,
+      "step": 605
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.995654455243204e-05,
+      "loss": 2.9417,
+      "step": 610
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.995568843452468e-05,
+      "loss": 2.8451,
+      "step": 615
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99548239692075e-05,
+      "loss": 2.9869,
+      "step": 620
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.995395115662498e-05,
+      "loss": 2.8953,
+      "step": 625
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.995306999692293e-05,
+      "loss": 3.0174,
+      "step": 630
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.995218049024858e-05,
+      "loss": 2.9253,
+      "step": 635
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.995128263675058e-05,
+      "loss": 2.8832,
+      "step": 640
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.995037643657899e-05,
+      "loss": 2.9136,
+      "step": 645
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994946188988516e-05,
+      "loss": 2.976,
+      "step": 650
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994853899682197e-05,
+      "loss": 2.9096,
+      "step": 655
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994760775754362e-05,
+      "loss": 2.9377,
+      "step": 660
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994666817220571e-05,
+      "loss": 2.8821,
+      "step": 665
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994572024096521e-05,
+      "loss": 2.8702,
+      "step": 670
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994476396398055e-05,
+      "loss": 2.8832,
+      "step": 675
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994379934141153e-05,
+      "loss": 2.8853,
+      "step": 680
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994282637341932e-05,
+      "loss": 2.9375,
+      "step": 685
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994184506016651e-05,
+      "loss": 2.9632,
+      "step": 690
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994085540181703e-05,
+      "loss": 2.9365,
+      "step": 695
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99398573985363e-05,
+      "loss": 2.812,
+      "step": 700
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993885105049107e-05,
+      "loss": 2.8884,
+      "step": 705
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993783635784947e-05,
+      "loss": 2.9183,
+      "step": 710
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993681332078108e-05,
+      "loss": 2.9664,
+      "step": 715
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993578193945685e-05,
+      "loss": 2.9854,
+      "step": 720
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99347422140491e-05,
+      "loss": 2.9162,
+      "step": 725
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993369414473157e-05,
+      "loss": 2.9043,
+      "step": 730
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993263773167939e-05,
+      "loss": 2.8744,
+      "step": 735
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993157297506907e-05,
+      "loss": 2.9389,
+      "step": 740
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993049987507856e-05,
+      "loss": 2.9573,
+      "step": 745
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992941843188715e-05,
+      "loss": 2.8547,
+      "step": 750
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992832864567554e-05,
+      "loss": 2.8576,
+      "step": 755
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992723051662584e-05,
+      "loss": 2.9174,
+      "step": 760
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992612404492153e-05,
+      "loss": 2.9312,
+      "step": 765
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992500923074752e-05,
+      "loss": 2.8674,
+      "step": 770
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992388607429006e-05,
+      "loss": 2.8292,
+      "step": 775
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992275457573685e-05,
+      "loss": 2.8824,
+      "step": 780
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992161473527695e-05,
+      "loss": 2.8335,
+      "step": 785
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992046655310083e-05,
+      "loss": 2.865,
+      "step": 790
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991931002940034e-05,
+      "loss": 2.8996,
+      "step": 795
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991814516436873e-05,
+      "loss": 2.8697,
+      "step": 800
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991697195820065e-05,
+      "loss": 2.73,
+      "step": 805
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991579041109212e-05,
+      "loss": 2.9004,
+      "step": 810
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991460052324061e-05,
+      "loss": 2.9012,
+      "step": 815
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991340229484491e-05,
+      "loss": 2.9,
+      "step": 820
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991219572610526e-05,
+      "loss": 2.8386,
+      "step": 825
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991098081722325e-05,
+      "loss": 2.9496,
+      "step": 830
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990975756840193e-05,
+      "loss": 2.8943,
+      "step": 835
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990852597984566e-05,
+      "loss": 2.879,
+      "step": 840
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990728605176025e-05,
+      "loss": 2.8844,
+      "step": 845
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990603778435288e-05,
+      "loss": 2.835,
+      "step": 850
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990478117783214e-05,
+      "loss": 2.8816,
+      "step": 855
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990351623240799e-05,
+      "loss": 2.7564,
+      "step": 860
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990224294829181e-05,
+      "loss": 2.9142,
+      "step": 865
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990096132569636e-05,
+      "loss": 2.9387,
+      "step": 870
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.98996713648358e-05,
+      "loss": 2.8097,
+      "step": 875
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989837306592567e-05,
+      "loss": 2.8123,
+      "step": 880
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989706642918291e-05,
+      "loss": 2.9317,
+      "step": 885
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989575145482583e-05,
+      "loss": 2.8998,
+      "step": 890
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989442814307422e-05,
+      "loss": 2.8298,
+      "step": 895
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989309649414914e-05,
+      "loss": 2.8156,
+      "step": 900
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989175650827314e-05,
+      "loss": 2.859,
+      "step": 905
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.98904081856701e-05,
+      "loss": 2.8473,
+      "step": 910
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988905152656534e-05,
+      "loss": 2.8927,
+      "step": 915
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988768653118554e-05,
+      "loss": 2.7602,
+      "step": 920
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988631319975881e-05,
+      "loss": 2.7786,
+      "step": 925
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988493153251459e-05,
+      "loss": 2.7522,
+      "step": 930
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988354152968379e-05,
+      "loss": 2.8016,
+      "step": 935
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988214319149865e-05,
+      "loss": 2.8596,
+      "step": 940
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988073651819281e-05,
+      "loss": 2.7941,
+      "step": 945
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987932151000137e-05,
+      "loss": 2.8229,
+      "step": 950
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987789816716073e-05,
+      "loss": 2.8147,
+      "step": 955
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987646648990876e-05,
+      "loss": 2.7683,
+      "step": 960
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.987502647848466e-05,
+      "loss": 2.8254,
+      "step": 965
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.987357813312905e-05,
+      "loss": 2.7513,
+      "step": 970
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.987212145408395e-05,
+      "loss": 2.7512,
+      "step": 975
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.987065644159278e-05,
+      "loss": 2.832,
+      "step": 980
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.986918309590031e-05,
+      "loss": 2.8109,
+      "step": 985
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.986770141725277e-05,
+      "loss": 2.803,
+      "step": 990
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.986621140589772e-05,
+      "loss": 2.8365,
+      "step": 995
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.986471306208412e-05,
+      "loss": 2.8577,
+      "step": 1000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.986320638606235e-05,
+      "loss": 2.7144,
+      "step": 1005
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.986169137808419e-05,
+      "loss": 2.7714,
+      "step": 1010
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.986016803840275e-05,
+      "loss": 2.8041,
+      "step": 1015
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.985863636727262e-05,
+      "loss": 2.7417,
+      "step": 1020
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.985709636494971e-05,
+      "loss": 2.7473,
+      "step": 1025
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.985554803169134e-05,
+      "loss": 2.8879,
+      "step": 1030
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.985399136775627e-05,
+      "loss": 2.7776,
+      "step": 1035
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.985242637340458e-05,
+      "loss": 2.8174,
+      "step": 1040
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.985085304889778e-05,
+      "loss": 2.7718,
+      "step": 1045
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984927139449877e-05,
+      "loss": 2.8064,
+      "step": 1050
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984768141047184e-05,
+      "loss": 2.7949,
+      "step": 1055
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984608309708266e-05,
+      "loss": 2.7933,
+      "step": 1060
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984447645459831e-05,
+      "loss": 2.7504,
+      "step": 1065
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984286148328728e-05,
+      "loss": 2.8228,
+      "step": 1070
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984123818341937e-05,
+      "loss": 2.7018,
+      "step": 1075
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983960655526587e-05,
+      "loss": 2.7676,
+      "step": 1080
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983796659909941e-05,
+      "loss": 2.6977,
+      "step": 1085
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983631831519401e-05,
+      "loss": 2.7481,
+      "step": 1090
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983466170382511e-05,
+      "loss": 2.7375,
+      "step": 1095
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983299676526949e-05,
+      "loss": 2.7451,
+      "step": 1100
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.98313234998054e-05,
+      "loss": 2.8372,
+      "step": 1105
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.982964190771242e-05,
+      "loss": 2.7805,
+      "step": 1110
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.98279519892715e-05,
+      "loss": 2.8206,
+      "step": 1115
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.982625374476508e-05,
+      "loss": 2.7444,
+      "step": 1120
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.98245471744769e-05,
+      "loss": 2.8018,
+      "step": 1125
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.982283227869211e-05,
+      "loss": 2.7771,
+      "step": 1130
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.982110905769729e-05,
+      "loss": 2.751,
+      "step": 1135
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.981937751178037e-05,
+      "loss": 2.7459,
+      "step": 1140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.981763764123067e-05,
+      "loss": 2.6663,
+      "step": 1145
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.981588944633893e-05,
+      "loss": 2.7572,
+      "step": 1150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.981413292739727e-05,
+      "loss": 2.6778,
+      "step": 1155
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.981236808469922e-05,
+      "loss": 2.7992,
+      "step": 1160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.981059491853964e-05,
+      "loss": 2.6654,
+      "step": 1165
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.980881342921482e-05,
+      "loss": 2.7882,
+      "step": 1170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.980702361702246e-05,
+      "loss": 2.7463,
+      "step": 1175
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.980522548226162e-05,
+      "loss": 2.7842,
+      "step": 1180
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.980341902523279e-05,
+      "loss": 2.7438,
+      "step": 1185
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.980160424623776e-05,
+      "loss": 2.7193,
+      "step": 1190
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.979978114557984e-05,
+      "loss": 2.7284,
+      "step": 1195
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.979794972356362e-05,
+      "loss": 2.759,
+      "step": 1200
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.979610998049516e-05,
+      "loss": 2.6419,
+      "step": 1205
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.979426191668182e-05,
+      "loss": 2.7664,
+      "step": 1210
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.979240553243246e-05,
+      "loss": 2.7537,
+      "step": 1215
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.979054082805726e-05,
+      "loss": 2.8019,
+      "step": 1220
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.978866780386777e-05,
+      "loss": 2.6809,
+      "step": 1225
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.978678646017701e-05,
+      "loss": 2.7964,
+      "step": 1230
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.978489679729934e-05,
+      "loss": 2.7779,
+      "step": 1235
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.978299881555048e-05,
+      "loss": 2.7292,
+      "step": 1240
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.97810925152476e-05,
+      "loss": 2.7403,
+      "step": 1245
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.977917789670925e-05,
+      "loss": 2.7334,
+      "step": 1250
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.977725496025535e-05,
+      "loss": 2.7486,
+      "step": 1255
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.977532370620718e-05,
+      "loss": 2.7059,
+      "step": 1260
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.977338413488748e-05,
+      "loss": 2.7238,
+      "step": 1265
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.977143624662035e-05,
+      "loss": 2.6948,
+      "step": 1270
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.976948004173125e-05,
+      "loss": 2.6302,
+      "step": 1275
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.976751552054706e-05,
+      "loss": 2.7523,
+      "step": 1280
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.976554268339607e-05,
+      "loss": 2.7502,
+      "step": 1285
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.97635615306079e-05,
+      "loss": 2.725,
+      "step": 1290
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.976157206251361e-05,
+      "loss": 2.7268,
+      "step": 1295
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.975957427944563e-05,
+      "loss": 2.7368,
+      "step": 1300
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.975756818173778e-05,
+      "loss": 2.7707,
+      "step": 1305
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.975555376972529e-05,
+      "loss": 2.7694,
+      "step": 1310
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.975353104374474e-05,
+      "loss": 2.683,
+      "step": 1315
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.975150000413411e-05,
+      "loss": 2.6893,
+      "step": 1320
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.97494606512328e-05,
+      "loss": 2.6814,
+      "step": 1325
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.974741298538158e-05,
+      "loss": 2.6776,
+      "step": 1330
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.974535700692257e-05,
+      "loss": 2.7155,
+      "step": 1335
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.974329271619938e-05,
+      "loss": 2.65,
+      "step": 1340
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.97412201135569e-05,
+      "loss": 2.7988,
+      "step": 1345
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.973913919934147e-05,
+      "loss": 2.7073,
+      "step": 1350
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.973704997390079e-05,
+      "loss": 2.7226,
+      "step": 1355
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.973495243758398e-05,
+      "loss": 2.7703,
+      "step": 1360
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.973284659074151e-05,
+      "loss": 2.7383,
+      "step": 1365
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.973073243372528e-05,
+      "loss": 2.6893,
+      "step": 1370
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.972860996688853e-05,
+      "loss": 2.7154,
+      "step": 1375
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.972647919058595e-05,
+      "loss": 2.805,
+      "step": 1380
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.972434010517358e-05,
+      "loss": 2.7125,
+      "step": 1385
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.972219271100882e-05,
+      "loss": 2.7327,
+      "step": 1390
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.972003700845052e-05,
+      "loss": 2.6328,
+      "step": 1395
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.971787299785888e-05,
+      "loss": 2.6987,
+      "step": 1400
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.97157006795955e-05,
+      "loss": 2.7298,
+      "step": 1405
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.971352005402338e-05,
+      "loss": 2.7601,
+      "step": 1410
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.971133112150688e-05,
+      "loss": 2.7172,
+      "step": 1415
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.970913388241177e-05,
+      "loss": 2.795,
+      "step": 1420
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.970692833710517e-05,
+      "loss": 2.7421,
+      "step": 1425
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.970471448595566e-05,
+      "loss": 2.7201,
+      "step": 1430
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.970249232933315e-05,
+      "loss": 2.6493,
+      "step": 1435
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.970026186760896e-05,
+      "loss": 2.712,
+      "step": 1440
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.969802310115577e-05,
+      "loss": 2.6268,
+      "step": 1445
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.96957760303477e-05,
+      "loss": 2.6469,
+      "step": 1450
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.969352065556022e-05,
+      "loss": 2.671,
+      "step": 1455
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.969125697717017e-05,
+      "loss": 2.709,
+      "step": 1460
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.968898499555582e-05,
+      "loss": 2.7191,
+      "step": 1465
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.968670471109682e-05,
+      "loss": 2.7134,
+      "step": 1470
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.968441612417418e-05,
+      "loss": 2.6793,
+      "step": 1475
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.968211923517032e-05,
+      "loss": 2.6117,
+      "step": 1480
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.967981404446905e-05,
+      "loss": 2.6366,
+      "step": 1485
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.967750055245555e-05,
+      "loss": 2.6552,
+      "step": 1490
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.967517875951638e-05,
+      "loss": 2.7197,
+      "step": 1495
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.967284866603955e-05,
+      "loss": 2.6182,
+      "step": 1500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.967051027241436e-05,
+      "loss": 2.6778,
+      "step": 1505
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.966816357903155e-05,
+      "loss": 2.7079,
+      "step": 1510
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.96658085862833e-05,
+      "loss": 2.7186,
+      "step": 1515
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.966344529456305e-05,
+      "loss": 2.698,
+      "step": 1520
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.966107370426573e-05,
+      "loss": 2.6967,
+      "step": 1525
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.965869381578765e-05,
+      "loss": 2.6555,
+      "step": 1530
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.965630562952643e-05,
+      "loss": 2.72,
+      "step": 1535
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.965390914588115e-05,
+      "loss": 2.6676,
+      "step": 1540
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.965150436525225e-05,
+      "loss": 2.6497,
+      "step": 1545
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.964909128804159e-05,
+      "loss": 2.634,
+      "step": 1550
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.964666991465234e-05,
+      "loss": 2.7437,
+      "step": 1555
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.964424024548914e-05,
+      "loss": 2.7039,
+      "step": 1560
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.964180228095793e-05,
+      "loss": 2.6633,
+      "step": 1565
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.963935602146612e-05,
+      "loss": 2.6636,
+      "step": 1570
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.963690146742248e-05,
+      "loss": 2.5802,
+      "step": 1575
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.963443861923715e-05,
+      "loss": 2.6553,
+      "step": 1580
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.963196747732164e-05,
+      "loss": 2.6899,
+      "step": 1585
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.96294880420889e-05,
+      "loss": 2.6667,
+      "step": 1590
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.96270003139532e-05,
+      "loss": 2.5968,
+      "step": 1595
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.962450429333026e-05,
+      "loss": 2.6101,
+      "step": 1600
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.962199998063715e-05,
+      "loss": 2.6219,
+      "step": 1605
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.961948737629231e-05,
+      "loss": 2.716,
+      "step": 1610
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.961696648071561e-05,
+      "loss": 2.6022,
+      "step": 1615
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.961443729432828e-05,
+      "loss": 2.6426,
+      "step": 1620
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.961189981755294e-05,
+      "loss": 2.618,
+      "step": 1625
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.960935405081357e-05,
+      "loss": 2.7158,
+      "step": 1630
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.960679999453559e-05,
+      "loss": 2.6329,
+      "step": 1635
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.960423764914575e-05,
+      "loss": 2.6386,
+      "step": 1640
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.960166701507221e-05,
+      "loss": 2.7342,
+      "step": 1645
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.959908809274453e-05,
+      "loss": 2.7261,
+      "step": 1650
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.959650088259362e-05,
+      "loss": 2.6792,
+      "step": 1655
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.959390538505181e-05,
+      "loss": 2.6253,
+      "step": 1660
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.959130160055279e-05,
+      "loss": 2.7578,
+      "step": 1665
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.958868952953163e-05,
+      "loss": 2.6406,
+      "step": 1670
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.95860691724248e-05,
+      "loss": 2.6676,
+      "step": 1675
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.958344052967018e-05,
+      "loss": 2.6226,
+      "step": 1680
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.958080360170698e-05,
+      "loss": 2.6393,
+      "step": 1685
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.957815838897585e-05,
+      "loss": 2.6257,
+      "step": 1690
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.957550489191874e-05,
+      "loss": 2.6676,
+      "step": 1695
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.957284311097908e-05,
+      "loss": 2.7567,
+      "step": 1700
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.957017304660164e-05,
+      "loss": 2.5816,
+      "step": 1705
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.956749469923258e-05,
+      "loss": 2.6413,
+      "step": 1710
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.956480806931942e-05,
+      "loss": 2.7035,
+      "step": 1715
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.956211315731112e-05,
+      "loss": 2.7134,
+      "step": 1720
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.955940996365796e-05,
+      "loss": 2.6849,
+      "step": 1725
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.955669848881166e-05,
+      "loss": 2.6685,
+      "step": 1730
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.955397873322527e-05,
+      "loss": 2.5903,
+      "step": 1735
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.955125069735327e-05,
+      "loss": 2.6535,
+      "step": 1740
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.95485143816515e-05,
+      "loss": 2.5887,
+      "step": 1745
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.95457697865772e-05,
+      "loss": 2.5927,
+      "step": 1750
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.954301691258894e-05,
+      "loss": 2.5885,
+      "step": 1755
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.954025576014674e-05,
+      "loss": 2.6686,
+      "step": 1760
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.953748632971201e-05,
+      "loss": 2.6649,
+      "step": 1765
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.953470862174748e-05,
+      "loss": 2.6313,
+      "step": 1770
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.953192263671728e-05,
+      "loss": 2.6076,
+      "step": 1775
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.952912837508697e-05,
+      "loss": 2.6005,
+      "step": 1780
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.952632583732343e-05,
+      "loss": 2.6989,
+      "step": 1785
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.952351502389498e-05,
+      "loss": 2.5452,
+      "step": 1790
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.95206959352713e-05,
+      "loss": 2.6002,
+      "step": 1795
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.951786857192341e-05,
+      "loss": 2.5832,
+      "step": 1800
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.951503293432378e-05,
+      "loss": 2.6349,
+      "step": 1805
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.951218902294625e-05,
+      "loss": 2.5542,
+      "step": 1810
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.9509336838266e-05,
+      "loss": 2.6508,
+      "step": 1815
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.950647638075963e-05,
+      "loss": 2.6018,
+      "step": 1820
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.950360765090511e-05,
+      "loss": 2.6055,
+      "step": 1825
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.950073064918179e-05,
+      "loss": 2.6329,
+      "step": 1830
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.949784537607042e-05,
+      "loss": 2.6442,
+      "step": 1835
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.949495183205311e-05,
+      "loss": 2.6761,
+      "step": 1840
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.949205001761334e-05,
+      "loss": 2.5496,
+      "step": 1845
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.948913993323603e-05,
+      "loss": 2.6338,
+      "step": 1850
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.948622157940744e-05,
+      "loss": 2.6429,
+      "step": 1855
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.948329495661517e-05,
+      "loss": 2.5688,
+      "step": 1860
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.94803600653483e-05,
+      "loss": 2.6465,
+      "step": 1865
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.947741690609722e-05,
+      "loss": 2.6997,
+      "step": 1870
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.94744654793537e-05,
+      "loss": 2.6346,
+      "step": 1875
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.947150578561095e-05,
+      "loss": 2.5635,
+      "step": 1880
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.946853782536351e-05,
+      "loss": 2.6024,
+      "step": 1885
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.946556159910732e-05,
+      "loss": 2.6388,
+      "step": 1890
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.946257710733966e-05,
+      "loss": 2.6561,
+      "step": 1895
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.945958435055927e-05,
+      "loss": 2.5488,
+      "step": 1900
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.945658332926622e-05,
+      "loss": 2.5699,
+      "step": 1905
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.945357404396197e-05,
+      "loss": 2.5755,
+      "step": 1910
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.945055649514934e-05,
+      "loss": 2.5786,
+      "step": 1915
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.944753068333256e-05,
+      "loss": 2.6153,
+      "step": 1920
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.944449660901725e-05,
+      "loss": 2.6003,
+      "step": 1925
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.944145427271036e-05,
+      "loss": 2.6586,
+      "step": 1930
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.94384036749203e-05,
+      "loss": 2.6812,
+      "step": 1935
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.943534481615677e-05,
+      "loss": 2.5721,
+      "step": 1940
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.943227769693091e-05,
+      "loss": 2.6051,
+      "step": 1945
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.942920231775524e-05,
+      "loss": 2.5333,
+      "step": 1950
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.942611867914363e-05,
+      "loss": 2.5655,
+      "step": 1955
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.942302678161133e-05,
+      "loss": 2.6195,
+      "step": 1960
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.9419926625675e-05,
+      "loss": 2.5592,
+      "step": 1965
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.941681821185265e-05,
+      "loss": 2.5941,
+      "step": 1970
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.941370154066371e-05,
+      "loss": 2.5925,
+      "step": 1975
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.941057661262895e-05,
+      "loss": 2.62,
+      "step": 1980
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.940744342827055e-05,
+      "loss": 2.6915,
+      "step": 1985
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.940430198811202e-05,
+      "loss": 2.6359,
+      "step": 1990
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.940115229267831e-05,
+      "loss": 2.549,
+      "step": 1995
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.939799434249571e-05,
+      "loss": 2.6021,
+      "step": 2000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.93948281380919e-05,
+      "loss": 2.5143,
+      "step": 2005
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.939165367999597e-05,
+      "loss": 2.6312,
+      "step": 2010
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.938847096873831e-05,
+      "loss": 2.5815,
+      "step": 2015
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.938528000485078e-05,
+      "loss": 2.5936,
+      "step": 2020
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.938208078886655e-05,
+      "loss": 2.5432,
+      "step": 2025
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.937887332132023e-05,
+      "loss": 2.6207,
+      "step": 2030
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.937565760274776e-05,
+      "loss": 2.5179,
+      "step": 2035
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.937243363368645e-05,
+      "loss": 2.6638,
+      "step": 2040
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.936920141467505e-05,
+      "loss": 2.5626,
+      "step": 2045
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.936596094625364e-05,
+      "loss": 2.6314,
+      "step": 2050
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.936271222896367e-05,
+      "loss": 2.6593,
+      "step": 2055
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.935945526334803e-05,
+      "loss": 2.6576,
+      "step": 2060
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.935619004995089e-05,
+      "loss": 2.5565,
+      "step": 2065
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.935291658931791e-05,
+      "loss": 2.5514,
+      "step": 2070
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.934963488199604e-05,
+      "loss": 2.5637,
+      "step": 2075
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.934634492853366e-05,
+      "loss": 2.5834,
+      "step": 2080
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.93430467294805e-05,
+      "loss": 2.5615,
+      "step": 2085
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.933974028538768e-05,
+      "loss": 2.6289,
+      "step": 2090
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.933642559680769e-05,
+      "loss": 2.5931,
+      "step": 2095
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.93331026642944e-05,
+      "loss": 2.5205,
+      "step": 2100
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.932977148840307e-05,
+      "loss": 2.5902,
+      "step": 2105
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.93264320696903e-05,
+      "loss": 2.5608,
+      "step": 2110
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.932308440871413e-05,
+      "loss": 2.6083,
+      "step": 2115
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.931972850603394e-05,
+      "loss": 2.5167,
+      "step": 2120
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.931636436221048e-05,
+      "loss": 2.5705,
+      "step": 2125
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.931299197780586e-05,
+      "loss": 2.5931,
+      "step": 2130
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.930961135338361e-05,
+      "loss": 2.5492,
+      "step": 2135
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.930622248950864e-05,
+      "loss": 2.5686,
+      "step": 2140
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.93028253867472e-05,
+      "loss": 2.5624,
+      "step": 2145
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.929942004566695e-05,
+      "loss": 2.5789,
+      "step": 2150
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.929600646683686e-05,
+      "loss": 2.6164,
+      "step": 2155
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.929258465082738e-05,
+      "loss": 2.5861,
+      "step": 2160
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.928915459821027e-05,
+      "loss": 2.6558,
+      "step": 2165
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.928571630955865e-05,
+      "loss": 2.5026,
+      "step": 2170
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.928226978544708e-05,
+      "loss": 2.5344,
+      "step": 2175
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.927881502645146e-05,
+      "loss": 2.5764,
+      "step": 2180
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.927535203314904e-05,
+      "loss": 2.5933,
+      "step": 2185
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.927188080611849e-05,
+      "loss": 2.6436,
+      "step": 2190
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.926840134593984e-05,
+      "loss": 2.6327,
+      "step": 2195
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.92649136531945e-05,
+      "loss": 2.615,
+      "step": 2200
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.926141772846525e-05,
+      "loss": 2.5767,
+      "step": 2205
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.925791357233623e-05,
+      "loss": 2.6786,
+      "step": 2210
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.925440118539298e-05,
+      "loss": 2.6727,
+      "step": 2215
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.925088056822241e-05,
+      "loss": 2.4989,
+      "step": 2220
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.924735172141281e-05,
+      "loss": 2.5128,
+      "step": 2225
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.924381464555381e-05,
+      "loss": 2.5849,
+      "step": 2230
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.924026934123647e-05,
+      "loss": 2.6144,
+      "step": 2235
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.92367158090532e-05,
+      "loss": 2.5669,
+      "step": 2240
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.923315404959775e-05,
+      "loss": 2.6299,
+      "step": 2245
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.92295840634653e-05,
+      "loss": 2.526,
+      "step": 2250
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.922600585125237e-05,
+      "loss": 2.5513,
+      "step": 2255
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.922241941355688e-05,
+      "loss": 2.5959,
+      "step": 2260
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.921882475097811e-05,
+      "loss": 2.567,
+      "step": 2265
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.921522186411671e-05,
+      "loss": 2.5548,
+      "step": 2270
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.92116107535747e-05,
+      "loss": 2.5991,
+      "step": 2275
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.920799141995549e-05,
+      "loss": 2.5868,
+      "step": 2280
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.920436386386387e-05,
+      "loss": 2.6032,
+      "step": 2285
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.920072808590596e-05,
+      "loss": 2.6004,
+      "step": 2290
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.91970840866893e-05,
+      "loss": 2.5341,
+      "step": 2295
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.919343186682282e-05,
+      "loss": 2.6481,
+      "step": 2300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.918977142691675e-05,
+      "loss": 2.5614,
+      "step": 2305
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.918610276758273e-05,
+      "loss": 2.5467,
+      "step": 2310
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.918242588943383e-05,
+      "loss": 2.6147,
+      "step": 2315
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.91787407930844e-05,
+      "loss": 2.5418,
+      "step": 2320
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.917504747915023e-05,
+      "loss": 2.6433,
+      "step": 2325
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.917134594824843e-05,
+      "loss": 2.4958,
+      "step": 2330
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.916763620099754e-05,
+      "loss": 2.5731,
+      "step": 2335
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.916391823801743e-05,
+      "loss": 2.5465,
+      "step": 2340
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.916019205992937e-05,
+      "loss": 2.5752,
+      "step": 2345
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.915645766735597e-05,
+      "loss": 2.6372,
+      "step": 2350
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.915271506092126e-05,
+      "loss": 2.5953,
+      "step": 2355
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.91489642412506e-05,
+      "loss": 2.5954,
+      "step": 2360
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.914520520897072e-05,
+      "loss": 2.4567,
+      "step": 2365
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.914143796470978e-05,
+      "loss": 2.6262,
+      "step": 2370
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.913766250909726e-05,
+      "loss": 2.5953,
+      "step": 2375
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.9133878842764e-05,
+      "loss": 2.5769,
+      "step": 2380
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.913008696634226e-05,
+      "loss": 2.5444,
+      "step": 2385
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.912628688046564e-05,
+      "loss": 2.5136,
+      "step": 2390
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.912247858576914e-05,
+      "loss": 2.5886,
+      "step": 2395
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.911866208288906e-05,
+      "loss": 2.5135,
+      "step": 2400
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.911483737246319e-05,
+      "loss": 2.5819,
+      "step": 2405
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.911100445513058e-05,
+      "loss": 2.5464,
+      "step": 2410
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.910716333153169e-05,
+      "loss": 2.5337,
+      "step": 2415
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.910331400230841e-05,
+      "loss": 2.6672,
+      "step": 2420
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.90994564681039e-05,
+      "loss": 2.5963,
+      "step": 2425
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.909559072956275e-05,
+      "loss": 2.5225,
+      "step": 2430
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.909171678733091e-05,
+      "loss": 2.5207,
+      "step": 2435
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.908783464205573e-05,
+      "loss": 2.3934,
+      "step": 2440
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.908394429438585e-05,
+      "loss": 2.5167,
+      "step": 2445
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.908004574497139e-05,
+      "loss": 2.5843,
+      "step": 2450
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.907613899446372e-05,
+      "loss": 2.5779,
+      "step": 2455
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.90722240435157e-05,
+      "loss": 2.5523,
+      "step": 2460
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.906830089278148e-05,
+      "loss": 2.6165,
+      "step": 2465
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.906436954291661e-05,
+      "loss": 2.6055,
+      "step": 2470
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.906042999457798e-05,
+      "loss": 2.4812,
+      "step": 2475
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.905648224842391e-05,
+      "loss": 2.5331,
+      "step": 2480
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.905252630511405e-05,
+      "loss": 2.6598,
+      "step": 2485
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.904856216530938e-05,
+      "loss": 2.5858,
+      "step": 2490
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.904458982967235e-05,
+      "loss": 2.5613,
+      "step": 2495
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.904060929886668e-05,
+      "loss": 2.5389,
+      "step": 2500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.903662057355751e-05,
+      "loss": 2.463,
+      "step": 2505
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.903262365441137e-05,
+      "loss": 2.5393,
+      "step": 2510
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.90286185420961e-05,
+      "loss": 2.5819,
+      "step": 2515
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.902460523728094e-05,
+      "loss": 2.5009,
+      "step": 2520
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.902058374063651e-05,
+      "loss": 2.5818,
+      "step": 2525
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.90165540528348e-05,
+      "loss": 2.5635,
+      "step": 2530
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.901251617454913e-05,
+      "loss": 2.6333,
+      "step": 2535
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.900847010645422e-05,
+      "loss": 2.5846,
+      "step": 2540
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.900441584922616e-05,
+      "loss": 2.5982,
+      "step": 2545
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.90003534035424e-05,
+      "loss": 2.5474,
+      "step": 2550
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.899628277008175e-05,
+      "loss": 2.5378,
+      "step": 2555
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.899220394952441e-05,
+      "loss": 2.5263,
+      "step": 2560
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.898811694255195e-05,
+      "loss": 2.5398,
+      "step": 2565
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.898402174984726e-05,
+      "loss": 2.437,
+      "step": 2570
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.897991837209464e-05,
+      "loss": 2.542,
+      "step": 2575
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.897580680997977e-05,
+      "loss": 2.5537,
+      "step": 2580
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.897168706418965e-05,
+      "loss": 2.5002,
+      "step": 2585
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.89675591354127e-05,
+      "loss": 2.5168,
+      "step": 2590
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.896342302433866e-05,
+      "loss": 2.4925,
+      "step": 2595
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.895927873165867e-05,
+      "loss": 2.6054,
+      "step": 2600
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.895512625806522e-05,
+      "loss": 2.5779,
+      "step": 2605
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.895096560425218e-05,
+      "loss": 2.5375,
+      "step": 2610
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.894679677091477e-05,
+      "loss": 2.7073,
+      "step": 2615
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.89426197587496e-05,
+      "loss": 2.5948,
+      "step": 2620
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.893843456845462e-05,
+      "loss": 2.5799,
+      "step": 2625
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.893424120072916e-05,
+      "loss": 2.4932,
+      "step": 2630
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.893003965627394e-05,
+      "loss": 2.5786,
+      "step": 2635
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.8925829935791e-05,
+      "loss": 2.4787,
+      "step": 2640
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.892161203998375e-05,
+      "loss": 2.4588,
+      "step": 2645
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.891738596955703e-05,
+      "loss": 2.5322,
+      "step": 2650
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.891315172521696e-05,
+      "loss": 2.5449,
+      "step": 2655
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.890890930767111e-05,
+      "loss": 2.5554,
+      "step": 2660
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.890465871762832e-05,
+      "loss": 2.6011,
+      "step": 2665
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.89003999557989e-05,
+      "loss": 2.4703,
+      "step": 2670
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.889613302289443e-05,
+      "loss": 2.5348,
+      "step": 2675
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.889185791962792e-05,
+      "loss": 2.593,
+      "step": 2680
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.888757464671372e-05,
+      "loss": 2.5893,
+      "step": 2685
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.888328320486755e-05,
+      "loss": 2.4888,
+      "step": 2690
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.88789835948065e-05,
+      "loss": 2.5706,
+      "step": 2695
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.887467581724901e-05,
+      "loss": 2.5214,
+      "step": 2700
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.887035987291488e-05,
+      "loss": 2.4915,
+      "step": 2705
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.88660357625253e-05,
+      "loss": 2.5589,
+      "step": 2710
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.886170348680284e-05,
+      "loss": 2.5324,
+      "step": 2715
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.885736304647139e-05,
+      "loss": 2.5634,
+      "step": 2720
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.885301444225619e-05,
+      "loss": 2.5885,
+      "step": 2725
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.884865767488391e-05,
+      "loss": 2.5481,
+      "step": 2730
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.884429274508256e-05,
+      "loss": 2.4417,
+      "step": 2735
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.883991965358146e-05,
+      "loss": 2.5403,
+      "step": 2740
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.883553840111136e-05,
+      "loss": 2.6417,
+      "step": 2745
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.883114898840437e-05,
+      "loss": 2.5032,
+      "step": 2750
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.882675141619391e-05,
+      "loss": 2.5023,
+      "step": 2755
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.882234568521482e-05,
+      "loss": 2.497,
+      "step": 2760
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.881793179620328e-05,
+      "loss": 2.551,
+      "step": 2765
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.881350974989682e-05,
+      "loss": 2.5104,
+      "step": 2770
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.880907954703437e-05,
+      "loss": 2.5626,
+      "step": 2775
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.880464118835619e-05,
+      "loss": 2.5146,
+      "step": 2780
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.880019467460391e-05,
+      "loss": 2.496,
+      "step": 2785
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.879574000652052e-05,
+      "loss": 2.5295,
+      "step": 2790
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.87912771848504e-05,
+      "loss": 2.6015,
+      "step": 2795
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.878680621033925e-05,
+      "loss": 2.5696,
+      "step": 2800
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.878232708373415e-05,
+      "loss": 2.5262,
+      "step": 2805
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.877783980578357e-05,
+      "loss": 2.611,
+      "step": 2810
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.877334437723728e-05,
+      "loss": 2.569,
+      "step": 2815
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.87688407988465e-05,
+      "loss": 2.5021,
+      "step": 2820
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.876432907136372e-05,
+      "loss": 2.4804,
+      "step": 2825
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.875980919554287e-05,
+      "loss": 2.3843,
+      "step": 2830
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.875528117213915e-05,
+      "loss": 2.5696,
+      "step": 2835
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.875074500190922e-05,
+      "loss": 2.5972,
+      "step": 2840
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.874620068561104e-05,
+      "loss": 2.4699,
+      "step": 2845
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.874164822400396e-05,
+      "loss": 2.5013,
+      "step": 2850
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.873708761784866e-05,
+      "loss": 2.5458,
+      "step": 2855
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.873251886790722e-05,
+      "loss": 2.5953,
+      "step": 2860
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.872794197494307e-05,
+      "loss": 2.5447,
+      "step": 2865
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.872335693972094e-05,
+      "loss": 2.492,
+      "step": 2870
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.871876376300702e-05,
+      "loss": 2.5056,
+      "step": 2875
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.871416244556881e-05,
+      "loss": 2.4814,
+      "step": 2880
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.870955298817517e-05,
+      "loss": 2.538,
+      "step": 2885
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.87049353915963e-05,
+      "loss": 2.6675,
+      "step": 2890
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.87003096566038e-05,
+      "loss": 2.5041,
+      "step": 2895
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.869567578397062e-05,
+      "loss": 2.4779,
+      "step": 2900
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.869103377447105e-05,
+      "loss": 2.4518,
+      "step": 2905
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.868638362888076e-05,
+      "loss": 2.4625,
+      "step": 2910
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.868172534797678e-05,
+      "loss": 2.494,
+      "step": 2915
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.867705893253747e-05,
+      "loss": 2.5393,
+      "step": 2920
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.867238438334259e-05,
+      "loss": 2.5552,
+      "step": 2925
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.866770170117324e-05,
+      "loss": 2.5758,
+      "step": 2930
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.866301088681185e-05,
+      "loss": 2.5243,
+      "step": 2935
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.865831194104227e-05,
+      "loss": 2.619,
+      "step": 2940
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.865360486464967e-05,
+      "loss": 2.6098,
+      "step": 2945
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.864888965842056e-05,
+      "loss": 2.5577,
+      "step": 2950
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.864416632314287e-05,
+      "loss": 2.5259,
+      "step": 2955
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.863943485960583e-05,
+      "loss": 2.5208,
+      "step": 2960
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.863469526860005e-05,
+      "loss": 2.573,
+      "step": 2965
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.862994755091752e-05,
+      "loss": 2.5166,
+      "step": 2970
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.862519170735154e-05,
+      "loss": 2.5827,
+      "step": 2975
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.86204277386968e-05,
+      "loss": 2.4528,
+      "step": 2980
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.861565564574935e-05,
+      "loss": 2.5101,
+      "step": 2985
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.861087542930658e-05,
+      "loss": 2.4653,
+      "step": 2990
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.860608709016726e-05,
+      "loss": 2.5236,
+      "step": 2995
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.860129062913147e-05,
+      "loss": 2.4473,
+      "step": 3000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.859648604700072e-05,
+      "loss": 2.4949,
+      "step": 3005
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.859167334457781e-05,
+      "loss": 2.5509,
+      "step": 3010
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.858685252266696e-05,
+      "loss": 2.499,
+      "step": 3015
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.858202358207367e-05,
+      "loss": 2.5254,
+      "step": 3020
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.857718652360486e-05,
+      "loss": 2.5672,
+      "step": 3025
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.857234134806878e-05,
+      "loss": 2.4457,
+      "step": 3030
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.856748805627503e-05,
+      "loss": 2.493,
+      "step": 3035
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.856262664903462e-05,
+      "loss": 2.4592,
+      "step": 3040
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.855775712715981e-05,
+      "loss": 2.4627,
+      "step": 3045
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.855287949146432e-05,
+      "loss": 2.4508,
+      "step": 3050
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.854799374276318e-05,
+      "loss": 2.4907,
+      "step": 3055
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.854309988187277e-05,
+      "loss": 2.4758,
+      "step": 3060
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.853819790961086e-05,
+      "loss": 2.4988,
+      "step": 3065
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.853328782679652e-05,
+      "loss": 2.5382,
+      "step": 3070
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.852836963425021e-05,
+      "loss": 2.4644,
+      "step": 3075
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.852344333279375e-05,
+      "loss": 2.4461,
+      "step": 3080
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.851850892325032e-05,
+      "loss": 2.4831,
+      "step": 3085
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.851356640644443e-05,
+      "loss": 2.5051,
+      "step": 3090
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.850861578320196e-05,
+      "loss": 2.5114,
+      "step": 3095
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.850365705435013e-05,
+      "loss": 2.5164,
+      "step": 3100
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.849869022071753e-05,
+      "loss": 2.5186,
+      "step": 3105
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.849371528313411e-05,
+      "loss": 2.4502,
+      "step": 3110
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.848873224243114e-05,
+      "loss": 2.4982,
+      "step": 3115
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.848374109944128e-05,
+      "loss": 2.5395,
+      "step": 3120
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.847874185499855e-05,
+      "loss": 2.5765,
+      "step": 3125
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.847373450993828e-05,
+      "loss": 2.4727,
+      "step": 3130
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.846871906509719e-05,
+      "loss": 2.4313,
+      "step": 3135
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.846369552131334e-05,
+      "loss": 2.4775,
+      "step": 3140
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.845866387942614e-05,
+      "loss": 2.5383,
+      "step": 3145
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.845362414027637e-05,
+      "loss": 2.4404,
+      "step": 3150
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.844857630470614e-05,
+      "loss": 2.5329,
+      "step": 3155
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.844352037355893e-05,
+      "loss": 2.5365,
+      "step": 3160
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.843845634767957e-05,
+      "loss": 2.5044,
+      "step": 3165
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.843338422791424e-05,
+      "loss": 2.5634,
+      "step": 3170
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.842830401511048e-05,
+      "loss": 2.4862,
+      "step": 3175
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.842321571011714e-05,
+      "loss": 2.5678,
+      "step": 3180
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.841811931378448e-05,
+      "loss": 2.5617,
+      "step": 3185
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.841301482696413e-05,
+      "loss": 2.4401,
+      "step": 3190
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.840790225050896e-05,
+      "loss": 2.4636,
+      "step": 3195
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.840278158527328e-05,
+      "loss": 2.4986,
+      "step": 3200
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.839765283211278e-05,
+      "loss": 2.4423,
+      "step": 3205
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.83925159918844e-05,
+      "loss": 2.4973,
+      "step": 3210
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.838737106544651e-05,
+      "loss": 2.4815,
+      "step": 3215
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.838221805365881e-05,
+      "loss": 2.4852,
+      "step": 3220
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.837705695738236e-05,
+      "loss": 2.4048,
+      "step": 3225
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.837188777747954e-05,
+      "loss": 2.4556,
+      "step": 3230
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.836671051481411e-05,
+      "loss": 2.4027,
+      "step": 3235
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.836152517025116e-05,
+      "loss": 2.5694,
+      "step": 3240
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.835633174465715e-05,
+      "loss": 2.4581,
+      "step": 3245
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.83511302388999e-05,
+      "loss": 2.5152,
+      "step": 3250
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.834592065384853e-05,
+      "loss": 2.5622,
+      "step": 3255
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.834070299037357e-05,
+      "loss": 2.496,
+      "step": 3260
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.833547724934687e-05,
+      "loss": 2.4858,
+      "step": 3265
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.833024343164159e-05,
+      "loss": 2.5565,
+      "step": 3270
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.832500153813234e-05,
+      "loss": 2.3936,
+      "step": 3275
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.831975156969498e-05,
+      "loss": 2.4849,
+      "step": 3280
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.831449352720677e-05,
+      "loss": 2.4645,
+      "step": 3285
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.830922741154631e-05,
+      "loss": 2.451,
+      "step": 3290
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.830395322359357e-05,
+      "loss": 2.449,
+      "step": 3295
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.829867096422983e-05,
+      "loss": 2.3989,
+      "step": 3300
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.829338063433773e-05,
+      "loss": 2.5079,
+      "step": 3305
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.828808223480126e-05,
+      "loss": 2.5639,
+      "step": 3310
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.828277576650578e-05,
+      "loss": 2.4949,
+      "step": 3315
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.827746123033796e-05,
+      "loss": 2.5544,
+      "step": 3320
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.827213862718587e-05,
+      "loss": 2.5874,
+      "step": 3325
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.826680795793887e-05,
+      "loss": 2.4452,
+      "step": 3330
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.826146922348773e-05,
+      "loss": 2.4889,
+      "step": 3335
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.825612242472448e-05,
+      "loss": 2.475,
+      "step": 3340
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.82507675625426e-05,
+      "loss": 2.4046,
+      "step": 3345
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.824540463783682e-05,
+      "loss": 2.5111,
+      "step": 3350
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.824003365150329e-05,
+      "loss": 2.5783,
+      "step": 3355
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.82346546044395e-05,
+      "loss": 2.5067,
+      "step": 3360
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.822926749754424e-05,
+      "loss": 2.5669,
+      "step": 3365
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.822387233171771e-05,
+      "loss": 2.429,
+      "step": 3370
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.821846910786136e-05,
+      "loss": 2.5069,
+      "step": 3375
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.821305782687812e-05,
+      "loss": 2.5122,
+      "step": 3380
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.820763848967215e-05,
+      "loss": 2.4883,
+      "step": 3385
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.820221109714902e-05,
+      "loss": 2.495,
+      "step": 3390
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.81967756502156e-05,
+      "loss": 2.4809,
+      "step": 3395
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.819133214978017e-05,
+      "loss": 2.486,
+      "step": 3400
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.818588059675229e-05,
+      "loss": 2.4934,
+      "step": 3405
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.818042099204292e-05,
+      "loss": 2.4195,
+      "step": 3410
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.817495333656433e-05,
+      "loss": 2.5354,
+      "step": 3415
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.816947763123013e-05,
+      "loss": 2.4944,
+      "step": 3420
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.816399387695529e-05,
+      "loss": 2.5044,
+      "step": 3425
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.815850207465616e-05,
+      "loss": 2.4828,
+      "step": 3430
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.815300222525035e-05,
+      "loss": 2.5743,
+      "step": 3435
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.814749432965691e-05,
+      "loss": 2.359,
+      "step": 3440
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.814197838879615e-05,
+      "loss": 2.4576,
+      "step": 3445
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.813645440358981e-05,
+      "loss": 2.4533,
+      "step": 3450
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.813092237496087e-05,
+      "loss": 2.414,
+      "step": 3455
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.812538230383376e-05,
+      "loss": 2.4286,
+      "step": 3460
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.811983419113418e-05,
+      "loss": 2.4341,
+      "step": 3465
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.81142780377892e-05,
+      "loss": 2.5425,
+      "step": 3470
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.810871384472724e-05,
+      "loss": 2.4353,
+      "step": 3475
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.810314161287806e-05,
+      "loss": 2.4921,
+      "step": 3480
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.809756134317276e-05,
+      "loss": 2.5466,
+      "step": 3485
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.809197303654376e-05,
+      "loss": 2.5184,
+      "step": 3490
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.808637669392487e-05,
+      "loss": 2.4427,
+      "step": 3495
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.808077231625119e-05,
+      "loss": 2.5591,
+      "step": 3500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.807515990445922e-05,
+      "loss": 2.3715,
+      "step": 3505
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.806953945948676e-05,
+      "loss": 2.4637,
+      "step": 3510
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.806391098227297e-05,
+      "loss": 2.4889,
+      "step": 3515
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.805827447375835e-05,
+      "loss": 2.45,
+      "step": 3520
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.805262993488473e-05,
+      "loss": 2.4558,
+      "step": 3525
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.804697736659532e-05,
+      "loss": 2.5073,
+      "step": 3530
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.804131676983458e-05,
+      "loss": 2.4547,
+      "step": 3535
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.803564814554844e-05,
+      "loss": 2.4747,
+      "step": 3540
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.802997149468408e-05,
+      "loss": 2.4823,
+      "step": 3545
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.802428681819003e-05,
+      "loss": 2.453,
+      "step": 3550
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.801859411701621e-05,
+      "loss": 2.5248,
+      "step": 3555
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.801289339211384e-05,
+      "loss": 2.3732,
+      "step": 3560
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.800718464443547e-05,
+      "loss": 2.4577,
+      "step": 3565
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.800146787493503e-05,
+      "loss": 2.4941,
+      "step": 3570
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.799574308456775e-05,
+      "loss": 2.4762,
+      "step": 3575
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.799001027429026e-05,
+      "loss": 2.5098,
+      "step": 3580
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.798426944506048e-05,
+      "loss": 2.5079,
+      "step": 3585
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.797852059783766e-05,
+      "loss": 2.4186,
+      "step": 3590
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.79727637335824e-05,
+      "loss": 2.4927,
+      "step": 3595
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.79669988532567e-05,
+      "loss": 2.4988,
+      "step": 3600
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.79612259578238e-05,
+      "loss": 2.3841,
+      "step": 3605
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.795544504824836e-05,
+      "loss": 2.3973,
+      "step": 3610
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.794965612549635e-05,
+      "loss": 2.5408,
+      "step": 3615
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.794385919053506e-05,
+      "loss": 2.4762,
+      "step": 3620
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.793805424433313e-05,
+      "loss": 2.4355,
+      "step": 3625
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.793224128786058e-05,
+      "loss": 2.3727,
+      "step": 3630
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.79264203220887e-05,
+      "loss": 2.4944,
+      "step": 3635
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.792059134799016e-05,
+      "loss": 2.5032,
+      "step": 3640
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.791475436653896e-05,
+      "loss": 2.4994,
+      "step": 3645
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.790890937871046e-05,
+      "loss": 2.4989,
+      "step": 3650
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.79030563854813e-05,
+      "loss": 2.3663,
+      "step": 3655
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.789719538782952e-05,
+      "loss": 2.4888,
+      "step": 3660
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.789132638673445e-05,
+      "loss": 2.4555,
+      "step": 3665
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.78854493831768e-05,
+      "loss": 2.5256,
+      "step": 3670
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.787956437813859e-05,
+      "loss": 2.4203,
+      "step": 3675
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.787367137260317e-05,
+      "loss": 2.4584,
+      "step": 3680
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.786777036755525e-05,
+      "loss": 2.3973,
+      "step": 3685
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.786186136398085e-05,
+      "loss": 2.5115,
+      "step": 3690
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.785594436286738e-05,
+      "loss": 2.4036,
+      "step": 3695
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.78500193652035e-05,
+      "loss": 2.5173,
+      "step": 3700
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.784408637197929e-05,
+      "loss": 2.5224,
+      "step": 3705
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.783814538418612e-05,
+      "loss": 2.4238,
+      "step": 3710
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.783219640281671e-05,
+      "loss": 2.4628,
+      "step": 3715
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.78262394288651e-05,
+      "loss": 2.5402,
+      "step": 3720
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.782027446332671e-05,
+      "loss": 2.4246,
+      "step": 3725
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.781430150719822e-05,
+      "loss": 2.4612,
+      "step": 3730
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.780832056147772e-05,
+      "loss": 2.4946,
+      "step": 3735
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.78023316271646e-05,
+      "loss": 2.517,
+      "step": 3740
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.779633470525957e-05,
+      "loss": 2.4418,
+      "step": 3745
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.77903297967647e-05,
+      "loss": 2.3206,
+      "step": 3750
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.778431690268341e-05,
+      "loss": 2.4462,
+      "step": 3755
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.777829602402043e-05,
+      "loss": 2.5931,
+      "step": 3760
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.777226716178178e-05,
+      "loss": 2.4543,
+      "step": 3765
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.776623031697492e-05,
+      "loss": 2.4491,
+      "step": 3770
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.776018549060855e-05,
+      "loss": 2.4044,
+      "step": 3775
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.775413268369274e-05,
+      "loss": 2.4607,
+      "step": 3780
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.77480718972389e-05,
+      "loss": 2.5749,
+      "step": 3785
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.774200313225976e-05,
+      "loss": 2.4482,
+      "step": 3790
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.773592638976939e-05,
+      "loss": 2.4633,
+      "step": 3795
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.77298416707832e-05,
+      "loss": 2.5084,
+      "step": 3800
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.772374897631791e-05,
+      "loss": 2.4715,
+      "step": 3805
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.771764830739159e-05,
+      "loss": 2.6208,
+      "step": 3810
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.771153966502365e-05,
+      "loss": 2.499,
+      "step": 3815
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.77054230502348e-05,
+      "loss": 2.3991,
+      "step": 3820
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.769929846404711e-05,
+      "loss": 2.4753,
+      "step": 3825
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.769316590748401e-05,
+      "loss": 2.4579,
+      "step": 3830
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.768702538157018e-05,
+      "loss": 2.4797,
+      "step": 3835
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.76808768873317e-05,
+      "loss": 2.4151,
+      "step": 3840
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.767472042579595e-05,
+      "loss": 2.5,
+      "step": 3845
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.766855599799168e-05,
+      "loss": 2.4484,
+      "step": 3850
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.76623836049489e-05,
+      "loss": 2.3268,
+      "step": 3855
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.765620324769903e-05,
+      "loss": 2.4613,
+      "step": 3860
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.765001492727476e-05,
+      "loss": 2.4763,
+      "step": 3865
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.764381864471017e-05,
+      "loss": 2.4106,
+      "step": 3870
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.76376144010406e-05,
+      "loss": 2.4869,
+      "step": 3875
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.763140219730278e-05,
+      "loss": 2.4248,
+      "step": 3880
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.762518203453472e-05,
+      "loss": 2.4326,
+      "step": 3885
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.761895391377581e-05,
+      "loss": 2.358,
+      "step": 3890
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.761271783606674e-05,
+      "loss": 2.4101,
+      "step": 3895
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.760647380244952e-05,
+      "loss": 2.4693,
+      "step": 3900
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.760022181396753e-05,
+      "loss": 2.4155,
+      "step": 3905
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.759396187166543e-05,
+      "loss": 2.4516,
+      "step": 3910
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.758769397658926e-05,
+      "loss": 2.513,
+      "step": 3915
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.758141812978634e-05,
+      "loss": 2.4679,
+      "step": 3920
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.757513433230534e-05,
+      "loss": 2.4367,
+      "step": 3925
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.756884258519627e-05,
+      "loss": 2.4652,
+      "step": 3930
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.756254288951046e-05,
+      "loss": 2.3882,
+      "step": 3935
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.755623524630055e-05,
+      "loss": 2.3998,
+      "step": 3940
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.754991965662055e-05,
+      "loss": 2.4212,
+      "step": 3945
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.754359612152574e-05,
+      "loss": 2.5224,
+      "step": 3950
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.753726464207278e-05,
+      "loss": 2.4828,
+      "step": 3955
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.753092521931963e-05,
+      "loss": 2.4583,
+      "step": 3960
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.752457785432558e-05,
+      "loss": 2.4379,
+      "step": 3965
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.751822254815126e-05,
+      "loss": 2.3866,
+      "step": 3970
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.751185930185861e-05,
+      "loss": 2.4288,
+      "step": 3975
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.750548811651091e-05,
+      "loss": 2.5055,
+      "step": 3980
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.749910899317276e-05,
+      "loss": 2.4499,
+      "step": 3985
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.74927219329101e-05,
+      "loss": 2.4138,
+      "step": 3990
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.748632693679016e-05,
+      "loss": 2.4469,
+      "step": 3995
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.747992400588153e-05,
+      "loss": 2.5015,
+      "step": 4000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.747351314125413e-05,
+      "loss": 2.4046,
+      "step": 4005
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.746709434397916e-05,
+      "loss": 2.41,
+      "step": 4010
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.746066761512921e-05,
+      "loss": 2.4054,
+      "step": 4015
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.745423295577815e-05,
+      "loss": 2.4586,
+      "step": 4020
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.744779036700118e-05,
+      "loss": 2.446,
+      "step": 4025
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.744133984987485e-05,
+      "loss": 2.4743,
+      "step": 4030
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.743488140547701e-05,
+      "loss": 2.4098,
+      "step": 4035
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.742841503488682e-05,
+      "loss": 2.493,
+      "step": 4040
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.742194073918482e-05,
+      "loss": 2.4427,
+      "step": 4045
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.741545851945282e-05,
+      "loss": 2.5501,
+      "step": 4050
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.740896837677398e-05,
+      "loss": 2.484,
+      "step": 4055
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.740247031223279e-05,
+      "loss": 2.56,
+      "step": 4060
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.739596432691502e-05,
+      "loss": 2.3759,
+      "step": 4065
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.738945042190785e-05,
+      "loss": 2.5365,
+      "step": 4070
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.738292859829968e-05,
+      "loss": 2.4526,
+      "step": 4075
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.73763988571803e-05,
+      "loss": 2.4342,
+      "step": 4080
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.73698611996408e-05,
+      "loss": 2.402,
+      "step": 4085
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.736331562677361e-05,
+      "loss": 2.4161,
+      "step": 4090
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.735676213967247e-05,
+      "loss": 2.4968,
+      "step": 4095
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.735020073943244e-05,
+      "loss": 2.3882,
+      "step": 4100
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.734363142714988e-05,
+      "loss": 2.4561,
+      "step": 4105
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.733705420392254e-05,
+      "loss": 2.3774,
+      "step": 4110
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.733046907084944e-05,
+      "loss": 2.4873,
+      "step": 4115
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.732387602903091e-05,
+      "loss": 2.4642,
+      "step": 4120
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.731727507956862e-05,
+      "loss": 2.4036,
+      "step": 4125
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.73106662235656e-05,
+      "loss": 2.4277,
+      "step": 4130
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.730404946212614e-05,
+      "loss": 2.364,
+      "step": 4135
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.729742479635589e-05,
+      "loss": 2.4157,
+      "step": 4140
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.729079222736177e-05,
+      "loss": 2.4479,
+      "step": 4145
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.728415175625211e-05,
+      "loss": 2.4418,
+      "step": 4150
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.727750338413647e-05,
+      "loss": 2.4597,
+      "step": 4155
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.727084711212579e-05,
+      "loss": 2.5372,
+      "step": 4160
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.726418294133229e-05,
+      "loss": 2.4971,
+      "step": 4165
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.725751087286954e-05,
+      "loss": 2.3523,
+      "step": 4170
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.725083090785242e-05,
+      "loss": 2.3862,
+      "step": 4175
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.724414304739712e-05,
+      "loss": 2.389,
+      "step": 4180
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.723744729262116e-05,
+      "loss": 2.4401,
+      "step": 4185
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.723074364464337e-05,
+      "loss": 2.4194,
+      "step": 4190
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.722403210458393e-05,
+      "loss": 2.4616,
+      "step": 4195
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.721731267356426e-05,
+      "loss": 2.4542,
+      "step": 4200
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.721058535270722e-05,
+      "loss": 2.4208,
+      "step": 4205
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.720385014313686e-05,
+      "loss": 2.4653,
+      "step": 4210
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.719710704597865e-05,
+      "loss": 2.5384,
+      "step": 4215
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.719035606235931e-05,
+      "loss": 2.4524,
+      "step": 4220
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.718359719340693e-05,
+      "loss": 2.4209,
+      "step": 4225
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.717683044025086e-05,
+      "loss": 2.4982,
+      "step": 4230
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.717005580402184e-05,
+      "loss": 2.4402,
+      "step": 4235
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.716327328585183e-05,
+      "loss": 2.4731,
+      "step": 4240
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.715648288687421e-05,
+      "loss": 2.4575,
+      "step": 4245
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.714968460822363e-05,
+      "loss": 2.4507,
+      "step": 4250
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.714287845103602e-05,
+      "loss": 2.4296,
+      "step": 4255
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.71360644164487e-05,
+      "loss": 2.5081,
+      "step": 4260
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.712924250560025e-05,
+      "loss": 2.5006,
+      "step": 4265
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.712241271963059e-05,
+      "loss": 2.4304,
+      "step": 4270
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.711557505968097e-05,
+      "loss": 2.4358,
+      "step": 4275
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.710872952689391e-05,
+      "loss": 2.4481,
+      "step": 4280
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.71018761224133e-05,
+      "loss": 2.4965,
+      "step": 4285
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.709501484738429e-05,
+      "loss": 2.4501,
+      "step": 4290
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.70881457029534e-05,
+      "loss": 2.4652,
+      "step": 4295
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.708126869026841e-05,
+      "loss": 2.3824,
+      "step": 4300
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.707438381047846e-05,
+      "loss": 2.4616,
+      "step": 4305
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.7067491064734e-05,
+      "loss": 2.4173,
+      "step": 4310
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.706059045418676e-05,
+      "loss": 2.4227,
+      "step": 4315
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.705368197998982e-05,
+      "loss": 2.3495,
+      "step": 4320
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.704676564329756e-05,
+      "loss": 2.5508,
+      "step": 4325
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.703984144526565e-05,
+      "loss": 2.4394,
+      "step": 4330
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.703290938705114e-05,
+      "loss": 2.4539,
+      "step": 4335
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.702596946981232e-05,
+      "loss": 2.4502,
+      "step": 4340
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.701902169470883e-05,
+      "loss": 2.511,
+      "step": 4345
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.701206606290164e-05,
+      "loss": 2.4004,
+      "step": 4350
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.700510257555295e-05,
+      "loss": 2.4589,
+      "step": 4355
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.69981312338264e-05,
+      "loss": 2.4317,
+      "step": 4360
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.699115203888686e-05,
+      "loss": 2.5044,
+      "step": 4365
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.69841649919005e-05,
+      "loss": 2.4204,
+      "step": 4370
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.697717009403485e-05,
+      "loss": 2.4217,
+      "step": 4375
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.697016734645873e-05,
+      "loss": 2.4887,
+      "step": 4380
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.696315675034227e-05,
+      "loss": 2.5072,
+      "step": 4385
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.69561383068569e-05,
+      "loss": 2.3779,
+      "step": 4390
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.694911201717541e-05,
+      "loss": 2.3973,
+      "step": 4395
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.694207788247186e-05,
+      "loss": 2.4722,
+      "step": 4400
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.693503590392159e-05,
+      "loss": 2.4574,
+      "step": 4405
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.692798608270132e-05,
+      "loss": 2.4608,
+      "step": 4410
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.692092841998904e-05,
+      "loss": 2.3768,
+      "step": 4415
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.691386291696407e-05,
+      "loss": 2.4286,
+      "step": 4420
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.690678957480702e-05,
+      "loss": 2.4277,
+      "step": 4425
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.689970839469981e-05,
+      "loss": 2.3301,
+      "step": 4430
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.689261937782571e-05,
+      "loss": 2.4029,
+      "step": 4435
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.688552252536924e-05,
+      "loss": 2.4598,
+      "step": 4440
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.687841783851626e-05,
+      "loss": 2.3926,
+      "step": 4445
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.687130531845394e-05,
+      "loss": 2.514,
+      "step": 4450
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.686418496637076e-05,
+      "loss": 2.4069,
+      "step": 4455
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.685705678345652e-05,
+      "loss": 2.4124,
+      "step": 4460
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.684992077090227e-05,
+      "loss": 2.4249,
+      "step": 4465
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.684277692990045e-05,
+      "loss": 2.4574,
+      "step": 4470
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.683562526164475e-05,
+      "loss": 2.4921,
+      "step": 4475
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.68284657673302e-05,
+      "loss": 2.3271,
+      "step": 4480
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.682129844815311e-05,
+      "loss": 2.415,
+      "step": 4485
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.681412330531112e-05,
+      "loss": 2.4421,
+      "step": 4490
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.680694034000318e-05,
+      "loss": 2.3808,
+      "step": 4495
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.67997495534295e-05,
+      "loss": 2.4039,
+      "step": 4500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.679255094679169e-05,
+      "loss": 2.4056,
+      "step": 4505
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.678534452129256e-05,
+      "loss": 2.3535,
+      "step": 4510
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.677813027813632e-05,
+      "loss": 2.4976,
+      "step": 4515
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.677090821852842e-05,
+      "loss": 2.4315,
+      "step": 4520
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.676367834367562e-05,
+      "loss": 2.4125,
+      "step": 4525
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.675644065478602e-05,
+      "loss": 2.4786,
+      "step": 4530
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.674919515306904e-05,
+      "loss": 2.3909,
+      "step": 4535
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.674194183973533e-05,
+      "loss": 2.4618,
+      "step": 4540
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.673468071599694e-05,
+      "loss": 2.4316,
+      "step": 4545
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.672741178306713e-05,
+      "loss": 2.468,
+      "step": 4550
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.672013504216054e-05,
+      "loss": 2.367,
+      "step": 4555
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.671285049449307e-05,
+      "loss": 2.4084,
+      "step": 4560
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.670555814128197e-05,
+      "loss": 2.3911,
+      "step": 4565
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.669825798374574e-05,
+      "loss": 2.3795,
+      "step": 4570
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.66909500231042e-05,
+      "loss": 2.5331,
+      "step": 4575
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.668363426057852e-05,
+      "loss": 2.402,
+      "step": 4580
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.66763106973911e-05,
+      "loss": 2.4982,
+      "step": 4585
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.666897933476568e-05,
+      "loss": 2.4059,
+      "step": 4590
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.666164017392734e-05,
+      "loss": 2.386,
+      "step": 4595
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.665429321610239e-05,
+      "loss": 2.4508,
+      "step": 4600
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.66469384625185e-05,
+      "loss": 2.4337,
+      "step": 4605
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.66395759144046e-05,
+      "loss": 2.487,
+      "step": 4610
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.6632205572991e-05,
+      "loss": 2.356,
+      "step": 4615
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.662482743950918e-05,
+      "loss": 2.3141,
+      "step": 4620
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.661744151519206e-05,
+      "loss": 2.4929,
+      "step": 4625
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.661004780127373e-05,
+      "loss": 2.3714,
+      "step": 4630
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.660264629898974e-05,
+      "loss": 2.5007,
+      "step": 4635
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.65952370095768e-05,
+      "loss": 2.4169,
+      "step": 4640
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.6587819934273e-05,
+      "loss": 2.4264,
+      "step": 4645
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.658039507431768e-05,
+      "loss": 2.44,
+      "step": 4650
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.657296243095151e-05,
+      "loss": 2.4165,
+      "step": 4655
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.656552200541648e-05,
+      "loss": 2.4213,
+      "step": 4660
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.655807379895583e-05,
+      "loss": 2.4216,
+      "step": 4665
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.655061781281413e-05,
+      "loss": 2.3755,
+      "step": 4670
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.654315404823729e-05,
+      "loss": 2.4617,
+      "step": 4675
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.653568250647242e-05,
+      "loss": 2.4198,
+      "step": 4680
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.652820318876802e-05,
+      "loss": 2.4181,
+      "step": 4685
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.652071609637385e-05,
+      "loss": 2.4922,
+      "step": 4690
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.651322123054097e-05,
+      "loss": 2.4271,
+      "step": 4695
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.650571859252175e-05,
+      "loss": 2.4396,
+      "step": 4700
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.649820818356985e-05,
+      "loss": 2.4953,
+      "step": 4705
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.649069000494025e-05,
+      "loss": 2.4008,
+      "step": 4710
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.648316405788916e-05,
+      "loss": 2.425,
+      "step": 4715
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.647563034367419e-05,
+      "loss": 2.3862,
+      "step": 4720
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.646808886355417e-05,
+      "loss": 2.3141,
+      "step": 4725
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.646053961878925e-05,
+      "loss": 2.4184,
+      "step": 4730
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.64529826106409e-05,
+      "loss": 2.5528,
+      "step": 4735
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.644541784037186e-05,
+      "loss": 2.4058,
+      "step": 4740
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.643784530924616e-05,
+      "loss": 2.4614,
+      "step": 4745
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.643026501852916e-05,
+      "loss": 2.4277,
+      "step": 4750
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.64226769694875e-05,
+      "loss": 2.3981,
+      "step": 4755
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.641508116338908e-05,
+      "loss": 2.3055,
+      "step": 4760
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.640747760150319e-05,
+      "loss": 2.3563,
+      "step": 4765
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.639986628510032e-05,
+      "loss": 2.4456,
+      "step": 4770
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.639224721545229e-05,
+      "loss": 2.4669,
+      "step": 4775
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.638462039383223e-05,
+      "loss": 2.3743,
+      "step": 4780
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.637698582151454e-05,
+      "loss": 2.4432,
+      "step": 4785
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.636934349977493e-05,
+      "loss": 2.3685,
+      "step": 4790
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.636169342989043e-05,
+      "loss": 2.5236,
+      "step": 4795
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.635403561313932e-05,
+      "loss": 2.443,
+      "step": 4800
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.634637005080119e-05,
+      "loss": 2.341,
+      "step": 4805
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.633869674415693e-05,
+      "loss": 2.3081,
+      "step": 4810
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.633101569448873e-05,
+      "loss": 2.4338,
+      "step": 4815
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.632332690308004e-05,
+      "loss": 2.4958,
+      "step": 4820
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.631563037121565e-05,
+      "loss": 2.4157,
+      "step": 4825
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.63079261001816e-05,
+      "loss": 2.4878,
+      "step": 4830
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.630021409126529e-05,
+      "loss": 2.4205,
+      "step": 4835
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.629249434575531e-05,
+      "loss": 2.473,
+      "step": 4840
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.628476686494164e-05,
+      "loss": 2.4482,
+      "step": 4845
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.627703165011551e-05,
+      "loss": 2.3596,
+      "step": 4850
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.626928870256944e-05,
+      "loss": 2.4343,
+      "step": 4855
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.626153802359724e-05,
+      "loss": 2.4824,
+      "step": 4860
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.625377961449402e-05,
+      "loss": 2.3757,
+      "step": 4865
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.62460134765562e-05,
+      "loss": 2.4572,
+      "step": 4870
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.623823961108145e-05,
+      "loss": 2.3549,
+      "step": 4875
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.623045801936876e-05,
+      "loss": 2.3924,
+      "step": 4880
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.622266870271842e-05,
+      "loss": 2.4595,
+      "step": 4885
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.6214871662432e-05,
+      "loss": 2.4235,
+      "step": 4890
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.620706689981232e-05,
+      "loss": 2.3941,
+      "step": 4895
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.619925441616354e-05,
+      "loss": 2.4906,
+      "step": 4900
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.619143421279112e-05,
+      "loss": 2.496,
+      "step": 4905
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.618360629100176e-05,
+      "loss": 2.4751,
+      "step": 4910
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.61757706521035e-05,
+      "loss": 2.4178,
+      "step": 4915
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.616792729740564e-05,
+      "loss": 2.4242,
+      "step": 4920
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.616007622821877e-05,
+      "loss": 2.4454,
+      "step": 4925
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.615221744585476e-05,
+      "loss": 2.4475,
+      "step": 4930
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.614435095162679e-05,
+      "loss": 2.3808,
+      "step": 4935
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.613647674684936e-05,
+      "loss": 2.4802,
+      "step": 4940
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.612859483283818e-05,
+      "loss": 2.3664,
+      "step": 4945
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.612070521091029e-05,
+      "loss": 2.4085,
+      "step": 4950
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.611280788238403e-05,
+      "loss": 2.4494,
+      "step": 4955
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.610490284857901e-05,
+      "loss": 2.4616,
+      "step": 4960
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.609699011081613e-05,
+      "loss": 2.5024,
+      "step": 4965
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.608906967041757e-05,
+      "loss": 2.52,
+      "step": 4970
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.608114152870684e-05,
+      "loss": 2.4155,
+      "step": 4975
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.607320568700866e-05,
+      "loss": 2.4875,
+      "step": 4980
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.606526214664911e-05,
+      "loss": 2.4337,
+      "step": 4985
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.605731090895551e-05,
+      "loss": 2.4352,
+      "step": 4990
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.604935197525649e-05,
+      "loss": 2.3665,
+      "step": 4995
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.604138534688196e-05,
+      "loss": 2.4805,
+      "step": 5000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.60334110251631e-05,
+      "loss": 2.376,
+      "step": 5005
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.602542901143241e-05,
+      "loss": 2.4785,
+      "step": 5010
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.601743930702363e-05,
+      "loss": 2.4245,
+      "step": 5015
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.600944191327184e-05,
+      "loss": 2.4307,
+      "step": 5020
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.600143683151335e-05,
+      "loss": 2.3899,
+      "step": 5025
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.59934240630858e-05,
+      "loss": 2.3953,
+      "step": 5030
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.598540360932806e-05,
+      "loss": 2.4595,
+      "step": 5035
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.597737547158035e-05,
+      "loss": 2.4115,
+      "step": 5040
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.596933965118413e-05,
+      "loss": 2.4223,
+      "step": 5045
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.596129614948217e-05,
+      "loss": 2.4095,
+      "step": 5050
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.595324496781849e-05,
+      "loss": 2.4695,
+      "step": 5055
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.594518610753841e-05,
+      "loss": 2.4347,
+      "step": 5060
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.593711956998855e-05,
+      "loss": 2.4148,
+      "step": 5065
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.59290453565168e-05,
+      "loss": 2.3755,
+      "step": 5070
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.592096346847232e-05,
+      "loss": 2.5178,
+      "step": 5075
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.591287390720555e-05,
+      "loss": 2.3826,
+      "step": 5080
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.590477667406828e-05,
+      "loss": 2.4134,
+      "step": 5085
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.589667177041347e-05,
+      "loss": 2.3611,
+      "step": 5090
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.588855919759544e-05,
+      "loss": 2.5018,
+      "step": 5095
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.588043895696978e-05,
+      "loss": 2.4999,
+      "step": 5100
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.587231104989334e-05,
+      "loss": 2.3415,
+      "step": 5105
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.586417547772426e-05,
+      "loss": 2.3298,
+      "step": 5110
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.585603224182196e-05,
+      "loss": 2.4523,
+      "step": 5115
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.584788134354718e-05,
+      "loss": 2.3998,
+      "step": 5120
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.583972278426186e-05,
+      "loss": 2.4079,
+      "step": 5125
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.583155656532929e-05,
+      "loss": 2.4353,
+      "step": 5130
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.582338268811402e-05,
+      "loss": 2.3627,
+      "step": 5135
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.581520115398185e-05,
+      "loss": 2.4149,
+      "step": 5140
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.580701196429991e-05,
+      "loss": 2.3867,
+      "step": 5145
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.579881512043655e-05,
+      "loss": 2.4297,
+      "step": 5150
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.579061062376146e-05,
+      "loss": 2.3547,
+      "step": 5155
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.578239847564559e-05,
+      "loss": 2.4153,
+      "step": 5160
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.577417867746113e-05,
+      "loss": 2.4862,
+      "step": 5165
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.576595123058158e-05,
+      "loss": 2.3847,
+      "step": 5170
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.575771613638175e-05,
+      "loss": 2.3136,
+      "step": 5175
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.574947339623765e-05,
+      "loss": 2.4036,
+      "step": 5180
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.574122301152665e-05,
+      "loss": 2.4091,
+      "step": 5185
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.573296498362733e-05,
+      "loss": 2.4312,
+      "step": 5190
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.572469931391959e-05,
+      "loss": 2.4346,
+      "step": 5195
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.571642600378459e-05,
+      "loss": 2.3648,
+      "step": 5200
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.570814505460475e-05,
+      "loss": 2.4257,
+      "step": 5205
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.569985646776382e-05,
+      "loss": 2.4415,
+      "step": 5210
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.569156024464677e-05,
+      "loss": 2.3568,
+      "step": 5215
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.568325638663989e-05,
+      "loss": 2.4136,
+      "step": 5220
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.567494489513068e-05,
+      "loss": 2.4942,
+      "step": 5225
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.5666625771508e-05,
+      "loss": 2.4303,
+      "step": 5230
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.565829901716194e-05,
+      "loss": 2.4921,
+      "step": 5235
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.564996463348384e-05,
+      "loss": 2.4995,
+      "step": 5240
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.564162262186638e-05,
+      "loss": 2.432,
+      "step": 5245
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.563327298370347e-05,
+      "loss": 2.3506,
+      "step": 5250
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.56249157203903e-05,
+      "loss": 2.3493,
+      "step": 5255
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.561655083332333e-05,
+      "loss": 2.2973,
+      "step": 5260
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.56081783239003e-05,
+      "loss": 2.4296,
+      "step": 5265
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.559979819352024e-05,
+      "loss": 2.4197,
+      "step": 5270
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.559141044358342e-05,
+      "loss": 2.3374,
+      "step": 5275
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.558301507549143e-05,
+      "loss": 2.3877,
+      "step": 5280
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.557461209064708e-05,
+      "loss": 2.4255,
+      "step": 5285
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.556620149045447e-05,
+      "loss": 2.4596,
+      "step": 5290
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.555778327631903e-05,
+      "loss": 2.3616,
+      "step": 5295
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.554935744964734e-05,
+      "loss": 2.3638,
+      "step": 5300
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.55409240118474e-05,
+      "loss": 2.3461,
+      "step": 5305
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.553248296432835e-05,
+      "loss": 2.2273,
+      "step": 5310
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.552403430850068e-05,
+      "loss": 2.4011,
+      "step": 5315
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.551557804577612e-05,
+      "loss": 2.3961,
+      "step": 5320
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.550711417756767e-05,
+      "loss": 2.3705,
+      "step": 5325
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.549864270528965e-05,
+      "loss": 2.3605,
+      "step": 5330
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.549016363035758e-05,
+      "loss": 2.3817,
+      "step": 5335
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.548167695418828e-05,
+      "loss": 2.42,
+      "step": 5340
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.547318267819986e-05,
+      "loss": 2.3337,
+      "step": 5345
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.546468080381167e-05,
+      "loss": 2.4398,
+      "step": 5350
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.545617133244436e-05,
+      "loss": 2.3508,
+      "step": 5355
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.544765426551981e-05,
+      "loss": 2.4143,
+      "step": 5360
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.543912960446117e-05,
+      "loss": 2.3485,
+      "step": 5365
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.543059735069293e-05,
+      "loss": 2.4093,
+      "step": 5370
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.542205750564077e-05,
+      "loss": 2.3629,
+      "step": 5375
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.541351007073167e-05,
+      "loss": 2.3558,
+      "step": 5380
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.540495504739387e-05,
+      "loss": 2.4293,
+      "step": 5385
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.53963924370569e-05,
+      "loss": 2.4217,
+      "step": 5390
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.538782224115151e-05,
+      "loss": 2.4395,
+      "step": 5395
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.537924446110976e-05,
+      "loss": 2.3819,
+      "step": 5400
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.537065909836497e-05,
+      "loss": 2.449,
+      "step": 5405
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.536206615435174e-05,
+      "loss": 2.4005,
+      "step": 5410
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.535346563050591e-05,
+      "loss": 2.3902,
+      "step": 5415
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.534485752826456e-05,
+      "loss": 2.3103,
+      "step": 5420
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.53362418490661e-05,
+      "loss": 2.5106,
+      "step": 5425
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.53276185943502e-05,
+      "loss": 2.4217,
+      "step": 5430
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.531898776555773e-05,
+      "loss": 2.3766,
+      "step": 5435
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.531034936413091e-05,
+      "loss": 2.4214,
+      "step": 5440
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.530170339151314e-05,
+      "loss": 2.5025,
+      "step": 5445
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.529304984914917e-05,
+      "loss": 2.3897,
+      "step": 5450
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.528438873848496e-05,
+      "loss": 2.4473,
+      "step": 5455
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.527572006096773e-05,
+      "loss": 2.3714,
+      "step": 5460
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.526704381804601e-05,
+      "loss": 2.4246,
+      "step": 5465
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.525836001116957e-05,
+      "loss": 2.4126,
+      "step": 5470
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.524966864178943e-05,
+      "loss": 2.4409,
+      "step": 5475
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.524096971135789e-05,
+      "loss": 2.4367,
+      "step": 5480
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.52322632213285e-05,
+      "loss": 2.3611,
+      "step": 5485
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.52235491731561e-05,
+      "loss": 2.3605,
+      "step": 5490
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.521482756829675e-05,
+      "loss": 2.3902,
+      "step": 5495
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.520609840820782e-05,
+      "loss": 2.3871,
+      "step": 5500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.51973616943479e-05,
+      "loss": 2.4253,
+      "step": 5505
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.518861742817689e-05,
+      "loss": 2.3491,
+      "step": 5510
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.51798656111559e-05,
+      "loss": 2.3555,
+      "step": 5515
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.517110624474733e-05,
+      "loss": 2.3684,
+      "step": 5520
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.516233933041485e-05,
+      "loss": 2.4407,
+      "step": 5525
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.515356486962337e-05,
+      "loss": 2.3983,
+      "step": 5530
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.514478286383907e-05,
+      "loss": 2.3584,
+      "step": 5535
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.513599331452939e-05,
+      "loss": 2.4625,
+      "step": 5540
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.512719622316304e-05,
+      "loss": 2.4337,
+      "step": 5545
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.511839159120997e-05,
+      "loss": 2.2443,
+      "step": 5550
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.51095794201414e-05,
+      "loss": 2.402,
+      "step": 5555
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.510075971142983e-05,
+      "loss": 2.3861,
+      "step": 5560
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.509193246654896e-05,
+      "loss": 2.4217,
+      "step": 5565
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.508309768697384e-05,
+      "loss": 2.3118,
+      "step": 5570
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.507425537418068e-05,
+      "loss": 2.4527,
+      "step": 5575
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.506540552964704e-05,
+      "loss": 2.4357,
+      "step": 5580
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.505654815485168e-05,
+      "loss": 2.3563,
+      "step": 5585
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.504768325127464e-05,
+      "loss": 2.3286,
+      "step": 5590
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.503881082039719e-05,
+      "loss": 2.4814,
+      "step": 5595
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.50299308637019e-05,
+      "loss": 2.4117,
+      "step": 5600
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.502104338267257e-05,
+      "loss": 2.4417,
+      "step": 5605
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.501214837879428e-05,
+      "loss": 2.3123,
+      "step": 5610
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.500324585355333e-05,
+      "loss": 2.4184,
+      "step": 5615
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.499433580843732e-05,
+      "loss": 2.3679,
+      "step": 5620
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.498541824493506e-05,
+      "loss": 2.4197,
+      "step": 5625
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.497649316453667e-05,
+      "loss": 2.3891,
+      "step": 5630
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.496756056873348e-05,
+      "loss": 2.3691,
+      "step": 5635
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.49586204590181e-05,
+      "loss": 2.4136,
+      "step": 5640
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.494967283688438e-05,
+      "loss": 2.3531,
+      "step": 5645
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.494071770382745e-05,
+      "loss": 2.3328,
+      "step": 5650
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.493175506134367e-05,
+      "loss": 2.3272,
+      "step": 5655
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.492278491093067e-05,
+      "loss": 2.4186,
+      "step": 5660
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.491380725408731e-05,
+      "loss": 2.4054,
+      "step": 5665
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.490482209231374e-05,
+      "loss": 2.51,
+      "step": 5670
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.489582942711135e-05,
+      "loss": 2.4327,
+      "step": 5675
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.488682925998277e-05,
+      "loss": 2.4759,
+      "step": 5680
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.48778215924319e-05,
+      "loss": 2.3678,
+      "step": 5685
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.48688064259639e-05,
+      "loss": 2.2701,
+      "step": 5690
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.485978376208513e-05,
+      "loss": 2.4447,
+      "step": 5695
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.485075360230328e-05,
+      "loss": 2.4545,
+      "step": 5700
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.484171594812728e-05,
+      "loss": 2.3471,
+      "step": 5705
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.483267080106723e-05,
+      "loss": 2.3884,
+      "step": 5710
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.482361816263456e-05,
+      "loss": 2.444,
+      "step": 5715
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.481455803434195e-05,
+      "loss": 2.3632,
+      "step": 5720
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.48054904177033e-05,
+      "loss": 2.4473,
+      "step": 5725
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.479641531423377e-05,
+      "loss": 2.2521,
+      "step": 5730
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.47873327254498e-05,
+      "loss": 2.2899,
+      "step": 5735
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.477824265286904e-05,
+      "loss": 2.3628,
+      "step": 5740
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.476914509801039e-05,
+      "loss": 2.4429,
+      "step": 5745
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.476004006239406e-05,
+      "loss": 2.3659,
+      "step": 5750
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.475092754754142e-05,
+      "loss": 2.3748,
+      "step": 5755
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.474180755497517e-05,
+      "loss": 2.3745,
+      "step": 5760
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.473268008621922e-05,
+      "loss": 2.4577,
+      "step": 5765
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.472354514279874e-05,
+      "loss": 2.378,
+      "step": 5770
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.471440272624012e-05,
+      "loss": 2.4122,
+      "step": 5775
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.470525283807105e-05,
+      "loss": 2.3767,
+      "step": 5780
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.469609547982044e-05,
+      "loss": 2.4826,
+      "step": 5785
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.468693065301843e-05,
+      "loss": 2.3814,
+      "step": 5790
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.467775835919646e-05,
+      "loss": 2.4169,
+      "step": 5795
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.466857859988715e-05,
+      "loss": 2.4579,
+      "step": 5800
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.465939137662444e-05,
+      "loss": 2.3573,
+      "step": 5805
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.465019669094346e-05,
+      "loss": 2.336,
+      "step": 5810
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.464099454438062e-05,
+      "loss": 2.3599,
+      "step": 5815
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.463178493847355e-05,
+      "loss": 2.482,
+      "step": 5820
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.462256787476113e-05,
+      "loss": 2.3187,
+      "step": 5825
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.461334335478354e-05,
+      "loss": 2.4525,
+      "step": 5830
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.460411138008214e-05,
+      "loss": 2.4113,
+      "step": 5835
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.459487195219954e-05,
+      "loss": 2.4533,
+      "step": 5840
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.458562507267963e-05,
+      "loss": 2.4205,
+      "step": 5845
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.457637074306754e-05,
+      "loss": 2.3707,
+      "step": 5850
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.456710896490961e-05,
+      "loss": 2.3933,
+      "step": 5855
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.455783973975348e-05,
+      "loss": 2.4705,
+      "step": 5860
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.454856306914797e-05,
+      "loss": 2.2709,
+      "step": 5865
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.453927895464321e-05,
+      "loss": 2.3342,
+      "step": 5870
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.452998739779051e-05,
+      "loss": 2.3927,
+      "step": 5875
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.452068840014246e-05,
+      "loss": 2.3154,
+      "step": 5880
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.451138196325291e-05,
+      "loss": 2.3812,
+      "step": 5885
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.45020680886769e-05,
+      "loss": 2.3231,
+      "step": 5890
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.449274677797078e-05,
+      "loss": 2.3984,
+      "step": 5895
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.448341803269208e-05,
+      "loss": 2.4197,
+      "step": 5900
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.44740818543996e-05,
+      "loss": 2.3821,
+      "step": 5905
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.44647382446534e-05,
+      "loss": 2.3647,
+      "step": 5910
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.445538720501474e-05,
+      "loss": 2.4243,
+      "step": 5915
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.444602873704616e-05,
+      "loss": 2.3606,
+      "step": 5920
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.443666284231141e-05,
+      "loss": 2.5041,
+      "step": 5925
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.44272895223755e-05,
+      "loss": 2.3639,
+      "step": 5930
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.441790877880469e-05,
+      "loss": 2.4169,
+      "step": 5935
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.440852061316645e-05,
+      "loss": 2.3524,
+      "step": 5940
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.439912502702952e-05,
+      "loss": 2.3777,
+      "step": 5945
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.438972202196387e-05,
+      "loss": 2.5213,
+      "step": 5950
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.43803115995407e-05,
+      "loss": 2.465,
+      "step": 5955
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.437089376133246e-05,
+      "loss": 2.3699,
+      "step": 5960
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.436146850891283e-05,
+      "loss": 2.353,
+      "step": 5965
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.435203584385673e-05,
+      "loss": 2.326,
+      "step": 5970
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.434259576774034e-05,
+      "loss": 2.3606,
+      "step": 5975
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.433314828214103e-05,
+      "loss": 2.3667,
+      "step": 5980
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.43236933886375e-05,
+      "loss": 2.3764,
+      "step": 5985
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.431423108880957e-05,
+      "loss": 2.4207,
+      "step": 5990
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.430476138423838e-05,
+      "loss": 2.3863,
+      "step": 5995
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.429528427650627e-05,
+      "loss": 2.501,
+      "step": 6000
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 38527,
+  "num_train_epochs": 1,
+  "save_steps": 1000,
+  "total_flos": 1.7134867590414336e+19,
+  "trial_name": null,
+  "trial_params": null
+}