diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,12768 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.999905824739841,
+  "eval_steps": 500,
+  "global_step": 10618,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999994528653223e-06,
+      "loss": 1.8507,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999978114624865e-06,
+      "loss": 1.4021,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999950757950849e-06,
+      "loss": 1.3668,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999912458691045e-06,
+      "loss": 1.3339,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999863216929276e-06,
+      "loss": 1.3698,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999803032773304e-06,
+      "loss": 1.3425,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999731906354849e-06,
+      "loss": 1.2789,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999649837829572e-06,
+      "loss": 1.3366,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999556827377083e-06,
+      "loss": 1.3464,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99945287520094e-06,
+      "loss": 1.3353,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999337981528646e-06,
+      "loss": 1.3226,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99921214661165e-06,
+      "loss": 1.2716,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999075370725346e-06,
+      "loss": 1.3135,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998927654169074e-06,
+      "loss": 1.3213,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998768997266117e-06,
+      "loss": 1.3275,
+      "step": 75
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.998599400363704e-06,
+      "loss": 1.3679,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.998418863832999e-06,
+      "loss": 1.2311,
+      "step": 85
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99822738806912e-06,
+      "loss": 1.3258,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.998024973491114e-06,
+      "loss": 1.2636,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.997811620541976e-06,
+      "loss": 1.3358,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.997587329688637e-06,
+      "loss": 1.236,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.997352101421962e-06,
+      "loss": 1.347,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.997105936256764e-06,
+      "loss": 1.3071,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99684883473178e-06,
+      "loss": 1.3087,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.996580797409687e-06,
+      "loss": 1.3576,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.996301824877099e-06,
+      "loss": 1.3194,
+      "step": 130
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.996011917744553e-06,
+      "loss": 1.3217,
+      "step": 135
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.995711076646526e-06,
+      "loss": 1.2644,
+      "step": 140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.995399302241418e-06,
+      "loss": 1.2433,
+      "step": 145
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99507659521156e-06,
+      "loss": 1.2806,
+      "step": 150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.994742956263208e-06,
+      "loss": 1.2965,
+      "step": 155
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.994398386126545e-06,
+      "loss": 1.3131,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.994042885555676e-06,
+      "loss": 1.2713,
+      "step": 165
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.993676455328629e-06,
+      "loss": 1.3255,
+      "step": 170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.993299096247348e-06,
+      "loss": 1.3215,
+      "step": 175
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.9929108091377e-06,
+      "loss": 1.2773,
+      "step": 180
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.992511594849465e-06,
+      "loss": 1.3232,
+      "step": 185
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99210145425634e-06,
+      "loss": 1.3358,
+      "step": 190
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.991680388255934e-06,
+      "loss": 1.3245,
+      "step": 195
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.991248397769763e-06,
+      "loss": 1.3153,
+      "step": 200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.990805483743258e-06,
+      "loss": 1.3301,
+      "step": 205
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.990351647145753e-06,
+      "loss": 1.3016,
+      "step": 210
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.989886888970488e-06,
+      "loss": 1.2853,
+      "step": 215
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.9894112102346e-06,
+      "loss": 1.2947,
+      "step": 220
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.988924611979137e-06,
+      "loss": 1.3119,
+      "step": 225
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.988427095269031e-06,
+      "loss": 1.2601,
+      "step": 230
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.987918661193121e-06,
+      "loss": 1.2447,
+      "step": 235
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987399310864135e-06,
+      "loss": 1.2965,
+      "step": 240
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986869045418689e-06,
+      "loss": 1.2928,
+      "step": 245
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98632786601729e-06,
+      "loss": 1.3183,
+      "step": 250
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98577577384433e-06,
+      "loss": 1.3156,
+      "step": 255
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985212770108087e-06,
+      "loss": 1.1941,
+      "step": 260
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984638856040712e-06,
+      "loss": 1.2811,
+      "step": 265
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98405403289824e-06,
+      "loss": 1.3099,
+      "step": 270
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98345830196058e-06,
+      "loss": 1.319,
+      "step": 275
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.982851664531511e-06,
+      "loss": 1.2916,
+      "step": 280
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.982234121938684e-06,
+      "loss": 1.2767,
+      "step": 285
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.981605675533612e-06,
+      "loss": 1.3185,
+      "step": 290
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980966326691677e-06,
+      "loss": 1.2588,
+      "step": 295
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.98031607681212e-06,
+      "loss": 1.2949,
+      "step": 300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979654927318033e-06,
+      "loss": 1.2925,
+      "step": 305
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97898287965637e-06,
+      "loss": 1.2809,
+      "step": 310
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978299935297935e-06,
+      "loss": 1.3291,
+      "step": 315
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977606095737375e-06,
+      "loss": 1.302,
+      "step": 320
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976901362493188e-06,
+      "loss": 1.2554,
+      "step": 325
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976185737107708e-06,
+      "loss": 1.2703,
+      "step": 330
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.975459221147108e-06,
+      "loss": 1.2658,
+      "step": 335
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.9747218162014e-06,
+      "loss": 1.2915,
+      "step": 340
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.973973523884419e-06,
+      "loss": 1.2699,
+      "step": 345
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.973214345833833e-06,
+      "loss": 1.3081,
+      "step": 350
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.972444283711133e-06,
+      "loss": 1.2531,
+      "step": 355
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.971663339201632e-06,
+      "loss": 1.2416,
+      "step": 360
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.970871514014453e-06,
+      "loss": 1.2025,
+      "step": 365
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.970068809882538e-06,
+      "loss": 1.3003,
+      "step": 370
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.969255228562637e-06,
+      "loss": 1.2742,
+      "step": 375
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.968430771835305e-06,
+      "loss": 1.3026,
+      "step": 380
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.967595441504894e-06,
+      "loss": 1.306,
+      "step": 385
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.966749239399558e-06,
+      "loss": 1.2697,
+      "step": 390
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.965892167371244e-06,
+      "loss": 1.3176,
+      "step": 395
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.965024227295688e-06,
+      "loss": 1.2822,
+      "step": 400
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.964145421072409e-06,
+      "loss": 1.2875,
+      "step": 405
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.96325575062471e-06,
+      "loss": 1.2609,
+      "step": 410
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.962355217899666e-06,
+      "loss": 1.2669,
+      "step": 415
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.961443824868131e-06,
+      "loss": 1.2581,
+      "step": 420
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.960521573524722e-06,
+      "loss": 1.2905,
+      "step": 425
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.959588465887824e-06,
+      "loss": 1.2999,
+      "step": 430
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.958644503999578e-06,
+      "loss": 1.3133,
+      "step": 435
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.95768968992588e-06,
+      "loss": 1.3436,
+      "step": 440
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.956724025756378e-06,
+      "loss": 1.2272,
+      "step": 445
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.955747513604466e-06,
+      "loss": 1.3054,
+      "step": 450
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.95476015560728e-06,
+      "loss": 1.2063,
+      "step": 455
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.953761953925687e-06,
+      "loss": 1.263,
+      "step": 460
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.952752910744295e-06,
+      "loss": 1.2657,
+      "step": 465
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.95173302827143e-06,
+      "loss": 1.2713,
+      "step": 470
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.950702308739149e-06,
+      "loss": 1.3124,
+      "step": 475
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.949660754403216e-06,
+      "loss": 1.303,
+      "step": 480
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.948608367543115e-06,
+      "loss": 1.2714,
+      "step": 485
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.94754515046204e-06,
+      "loss": 1.2749,
+      "step": 490
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.946471105486874e-06,
+      "loss": 1.281,
+      "step": 495
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.945386234968213e-06,
+      "loss": 1.2042,
+      "step": 500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.944290541280337e-06,
+      "loss": 1.2807,
+      "step": 505
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.943184026821211e-06,
+      "loss": 1.2698,
+      "step": 510
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.94206669401249e-06,
+      "loss": 1.2716,
+      "step": 515
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.940938545299492e-06,
+      "loss": 1.2998,
+      "step": 520
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.939799583151222e-06,
+      "loss": 1.2695,
+      "step": 525
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.93864981006034e-06,
+      "loss": 1.2706,
+      "step": 530
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.937489228543166e-06,
+      "loss": 1.2435,
+      "step": 535
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.936317841139682e-06,
+      "loss": 1.2424,
+      "step": 540
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.935135650413512e-06,
+      "loss": 1.2579,
+      "step": 545
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.933942658951927e-06,
+      "loss": 1.3081,
+      "step": 550
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.932738869365836e-06,
+      "loss": 1.2869,
+      "step": 555
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.931524284289777e-06,
+      "loss": 1.2493,
+      "step": 560
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.930298906381918e-06,
+      "loss": 1.2675,
+      "step": 565
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.929062738324047e-06,
+      "loss": 1.2846,
+      "step": 570
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.927815782821561e-06,
+      "loss": 1.2714,
+      "step": 575
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.926558042603476e-06,
+      "loss": 1.2439,
+      "step": 580
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.925289520422402e-06,
+      "loss": 1.2844,
+      "step": 585
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.924010219054552e-06,
+      "loss": 1.2298,
+      "step": 590
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.922720141299721e-06,
+      "loss": 1.2103,
+      "step": 595
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.921419289981297e-06,
+      "loss": 1.2766,
+      "step": 600
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.920107667946246e-06,
+      "loss": 1.2139,
+      "step": 605
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.918785278065102e-06,
+      "loss": 1.2914,
+      "step": 610
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.917452123231964e-06,
+      "loss": 1.2117,
+      "step": 615
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.916108206364495e-06,
+      "loss": 1.2564,
+      "step": 620
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.914753530403909e-06,
+      "loss": 1.2346,
+      "step": 625
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.913388098314966e-06,
+      "loss": 1.2285,
+      "step": 630
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.91201191308597e-06,
+      "loss": 1.2825,
+      "step": 635
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.91062497772875e-06,
+      "loss": 1.2464,
+      "step": 640
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.909227295278674e-06,
+      "loss": 1.2736,
+      "step": 645
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.907818868794619e-06,
+      "loss": 1.3481,
+      "step": 650
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.906399701358984e-06,
+      "loss": 1.3355,
+      "step": 655
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.904969796077672e-06,
+      "loss": 1.2001,
+      "step": 660
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.903529156080083e-06,
+      "loss": 1.2797,
+      "step": 665
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.902077784519117e-06,
+      "loss": 1.2513,
+      "step": 670
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.900615684571155e-06,
+      "loss": 1.2671,
+      "step": 675
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.899142859436057e-06,
+      "loss": 1.241,
+      "step": 680
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.897659312337163e-06,
+      "loss": 1.2976,
+      "step": 685
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.896165046521272e-06,
+      "loss": 1.2553,
+      "step": 690
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.894660065258638e-06,
+      "loss": 1.2546,
+      "step": 695
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.893144371842975e-06,
+      "loss": 1.2879,
+      "step": 700
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.891617969591435e-06,
+      "loss": 1.2645,
+      "step": 705
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.890080861844612e-06,
+      "loss": 1.2744,
+      "step": 710
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.88853305196652e-06,
+      "loss": 1.2994,
+      "step": 715
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.886974543344604e-06,
+      "loss": 1.2771,
+      "step": 720
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.885405339389721e-06,
+      "loss": 1.2538,
+      "step": 725
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.883825443536133e-06,
+      "loss": 1.2017,
+      "step": 730
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.882234859241506e-06,
+      "loss": 1.2562,
+      "step": 735
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.880633589986891e-06,
+      "loss": 1.305,
+      "step": 740
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.879021639276732e-06,
+      "loss": 1.2444,
+      "step": 745
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.877399010638842e-06,
+      "loss": 1.2616,
+      "step": 750
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.875765707624408e-06,
+      "loss": 1.3,
+      "step": 755
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.874121733807978e-06,
+      "loss": 1.2517,
+      "step": 760
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.872467092787452e-06,
+      "loss": 1.2363,
+      "step": 765
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.870801788184074e-06,
+      "loss": 1.2549,
+      "step": 770
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.86912582364243e-06,
+      "loss": 1.2361,
+      "step": 775
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.867439202830431e-06,
+      "loss": 1.2699,
+      "step": 780
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.865741929439314e-06,
+      "loss": 1.2812,
+      "step": 785
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.864034007183628e-06,
+      "loss": 1.2381,
+      "step": 790
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.862315439801224e-06,
+      "loss": 1.2707,
+      "step": 795
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.860586231053254e-06,
+      "loss": 1.2527,
+      "step": 800
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.85884638472416e-06,
+      "loss": 1.2304,
+      "step": 805
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.857095904621662e-06,
+      "loss": 1.2258,
+      "step": 810
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.855334794576756e-06,
+      "loss": 1.2851,
+      "step": 815
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.853563058443697e-06,
+      "loss": 1.2311,
+      "step": 820
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.851780700099995e-06,
+      "loss": 1.245,
+      "step": 825
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.849987723446415e-06,
+      "loss": 1.2513,
+      "step": 830
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.848184132406955e-06,
+      "loss": 1.2234,
+      "step": 835
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.846369930928841e-06,
+      "loss": 1.2398,
+      "step": 840
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.844545122982528e-06,
+      "loss": 1.274,
+      "step": 845
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.842709712561675e-06,
+      "loss": 1.2703,
+      "step": 850
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.84086370368315e-06,
+      "loss": 1.2394,
+      "step": 855
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.839007100387015e-06,
+      "loss": 1.2349,
+      "step": 860
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.837139906736516e-06,
+      "loss": 1.2681,
+      "step": 865
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.835262126818084e-06,
+      "loss": 1.2424,
+      "step": 870
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.833373764741307e-06,
+      "loss": 1.2286,
+      "step": 875
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.831474824638942e-06,
+      "loss": 1.2393,
+      "step": 880
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.829565310666893e-06,
+      "loss": 1.2585,
+      "step": 885
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.827645227004203e-06,
+      "loss": 1.1964,
+      "step": 890
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.825714577853052e-06,
+      "loss": 1.2804,
+      "step": 895
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.823773367438737e-06,
+      "loss": 1.2708,
+      "step": 900
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.821821600009675e-06,
+      "loss": 1.2707,
+      "step": 905
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.819859279837385e-06,
+      "loss": 1.2573,
+      "step": 910
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.81788641121648e-06,
+      "loss": 1.2277,
+      "step": 915
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.815902998464656e-06,
+      "loss": 1.2458,
+      "step": 920
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.813909045922693e-06,
+      "loss": 1.2527,
+      "step": 925
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.811904557954432e-06,
+      "loss": 1.2622,
+      "step": 930
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.80988953894677e-06,
+      "loss": 1.2964,
+      "step": 935
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.807863993309659e-06,
+      "loss": 1.2748,
+      "step": 940
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.805827925476081e-06,
+      "loss": 1.2746,
+      "step": 945
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.80378133990205e-06,
+      "loss": 1.2681,
+      "step": 950
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.801724241066597e-06,
+      "loss": 1.2258,
+      "step": 955
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.799656633471762e-06,
+      "loss": 1.2985,
+      "step": 960
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.797578521642585e-06,
+      "loss": 1.2894,
+      "step": 965
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.795489910127096e-06,
+      "loss": 1.254,
+      "step": 970
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.793390803496299e-06,
+      "loss": 1.2879,
+      "step": 975
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.791281206344172e-06,
+      "loss": 1.2504,
+      "step": 980
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.789161123287649e-06,
+      "loss": 1.2561,
+      "step": 985
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.787030558966615e-06,
+      "loss": 1.2503,
+      "step": 990
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.784889518043892e-06,
+      "loss": 1.2776,
+      "step": 995
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.782738005205232e-06,
+      "loss": 1.2433,
+      "step": 1000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.780576025159301e-06,
+      "loss": 1.2934,
+      "step": 1005
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.778403582637679e-06,
+      "loss": 1.2654,
+      "step": 1010
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.776220682394841e-06,
+      "loss": 1.2501,
+      "step": 1015
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.774027329208144e-06,
+      "loss": 1.2686,
+      "step": 1020
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.771823527877832e-06,
+      "loss": 1.2319,
+      "step": 1025
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.769609283227006e-06,
+      "loss": 1.2093,
+      "step": 1030
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.767384600101629e-06,
+      "loss": 1.2621,
+      "step": 1035
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.765149483370503e-06,
+      "loss": 1.2476,
+      "step": 1040
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.76290393792527e-06,
+      "loss": 1.2902,
+      "step": 1045
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.76064796868039e-06,
+      "loss": 1.2564,
+      "step": 1050
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.758381580573142e-06,
+      "loss": 1.2317,
+      "step": 1055
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.756104778563604e-06,
+      "loss": 1.2248,
+      "step": 1060
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.753817567634645e-06,
+      "loss": 1.2295,
+      "step": 1065
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.751519952791912e-06,
+      "loss": 1.2673,
+      "step": 1070
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.749211939063827e-06,
+      "loss": 1.2297,
+      "step": 1075
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.746893531501567e-06,
+      "loss": 1.2802,
+      "step": 1080
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.744564735179058e-06,
+      "loss": 1.2593,
+      "step": 1085
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.742225555192956e-06,
+      "loss": 1.1905,
+      "step": 1090
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.739875996662652e-06,
+      "loss": 1.2163,
+      "step": 1095
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.737516064730244e-06,
+      "loss": 1.1864,
+      "step": 1100
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.735145764560535e-06,
+      "loss": 1.3442,
+      "step": 1105
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.732765101341014e-06,
+      "loss": 1.302,
+      "step": 1110
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.730374080281864e-06,
+      "loss": 1.1606,
+      "step": 1115
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.727972706615917e-06,
+      "loss": 1.2689,
+      "step": 1120
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.72556098559868e-06,
+      "loss": 1.2357,
+      "step": 1125
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.723138922508295e-06,
+      "loss": 1.2881,
+      "step": 1130
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.72070652264554e-06,
+      "loss": 1.2428,
+      "step": 1135
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.718263791333816e-06,
+      "loss": 1.2177,
+      "step": 1140
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.715810733919138e-06,
+      "loss": 1.2671,
+      "step": 1145
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.713347355770114e-06,
+      "loss": 1.2208,
+      "step": 1150
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.710873662277943e-06,
+      "loss": 1.2466,
+      "step": 1155
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.708389658856399e-06,
+      "loss": 1.2036,
+      "step": 1160
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.70589535094182e-06,
+      "loss": 1.2119,
+      "step": 1165
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.703390743993095e-06,
+      "loss": 1.2693,
+      "step": 1170
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.700875843491655e-06,
+      "loss": 1.2261,
+      "step": 1175
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.698350654941457e-06,
+      "loss": 1.3165,
+      "step": 1180
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.695815183868971e-06,
+      "loss": 1.2342,
+      "step": 1185
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.693269435823176e-06,
+      "loss": 1.1965,
+      "step": 1190
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.69071341637554e-06,
+      "loss": 1.312,
+      "step": 1195
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.688147131120009e-06,
+      "loss": 1.2269,
+      "step": 1200
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.685570585672999e-06,
+      "loss": 1.297,
+      "step": 1205
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.68298378567338e-06,
+      "loss": 1.2408,
+      "step": 1210
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.680386736782463e-06,
+      "loss": 1.2226,
+      "step": 1215
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.677779444683988e-06,
+      "loss": 1.2508,
+      "step": 1220
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.67516191508412e-06,
+      "loss": 1.2752,
+      "step": 1225
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.672534153711417e-06,
+      "loss": 1.222,
+      "step": 1230
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.669896166316843e-06,
+      "loss": 1.306,
+      "step": 1235
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.667247958673731e-06,
+      "loss": 1.2811,
+      "step": 1240
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.664589536577788e-06,
+      "loss": 1.3114,
+      "step": 1245
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.661920905847072e-06,
+      "loss": 1.2425,
+      "step": 1250
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.659242072321988e-06,
+      "loss": 1.2564,
+      "step": 1255
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.656553041865264e-06,
+      "loss": 1.2775,
+      "step": 1260
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.653853820361949e-06,
+      "loss": 1.2803,
+      "step": 1265
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.651144413719393e-06,
+      "loss": 1.2335,
+      "step": 1270
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.648424827867236e-06,
+      "loss": 1.2067,
+      "step": 1275
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.645695068757398e-06,
+      "loss": 1.2266,
+      "step": 1280
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.64295514236406e-06,
+      "loss": 1.263,
+      "step": 1285
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.640205054683663e-06,
+      "loss": 1.2369,
+      "step": 1290
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.637444811734876e-06,
+      "loss": 1.2629,
+      "step": 1295
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.634674419558599e-06,
+      "loss": 1.1917,
+      "step": 1300
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.63189388421794e-06,
+      "loss": 1.2513,
+      "step": 1305
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.629103211798212e-06,
+      "loss": 1.2235,
+      "step": 1310
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.626302408406907e-06,
+      "loss": 1.2851,
+      "step": 1315
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.623491480173692e-06,
+      "loss": 1.2207,
+      "step": 1320
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.620670433250394e-06,
+      "loss": 1.2435,
+      "step": 1325
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.61783927381098e-06,
+      "loss": 1.272,
+      "step": 1330
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.614998008051554e-06,
+      "loss": 1.2395,
+      "step": 1335
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.612146642190338e-06,
+      "loss": 1.2018,
+      "step": 1340
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.609285182467653e-06,
+      "loss": 1.2505,
+      "step": 1345
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.606413635145918e-06,
+      "loss": 1.2256,
+      "step": 1350
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.603532006509622e-06,
+      "loss": 1.2002,
+      "step": 1355
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.600640302865325e-06,
+      "loss": 1.2226,
+      "step": 1360
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.597738530541628e-06,
+      "loss": 1.2916,
+      "step": 1365
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.594826695889172e-06,
+      "loss": 1.256,
+      "step": 1370
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.591904805280623e-06,
+      "loss": 1.1941,
+      "step": 1375
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.58897286511065e-06,
+      "loss": 1.2103,
+      "step": 1380
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.58603088179592e-06,
+      "loss": 1.2854,
+      "step": 1385
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.583078861775074e-06,
+      "loss": 1.2441,
+      "step": 1390
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.580116811508725e-06,
+      "loss": 1.252,
+      "step": 1395
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.57714473747943e-06,
+      "loss": 1.2738,
+      "step": 1400
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.574162646191693e-06,
+      "loss": 1.1774,
+      "step": 1405
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.571170544171933e-06,
+      "loss": 1.3026,
+      "step": 1410
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.568168437968484e-06,
+      "loss": 1.2147,
+      "step": 1415
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.56515633415157e-06,
+      "loss": 1.2113,
+      "step": 1420
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.562134239313297e-06,
+      "loss": 1.2178,
+      "step": 1425
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.559102160067635e-06,
+      "loss": 1.2427,
+      "step": 1430
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.55606010305041e-06,
+      "loss": 1.2265,
+      "step": 1435
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.553008074919278e-06,
+      "loss": 1.2229,
+      "step": 1440
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.549946082353724e-06,
+      "loss": 1.2404,
+      "step": 1445
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.546874132055036e-06,
+      "loss": 1.2487,
+      "step": 1450
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.543792230746294e-06,
+      "loss": 1.1743,
+      "step": 1455
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.540700385172363e-06,
+      "loss": 1.2155,
+      "step": 1460
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.537598602099862e-06,
+      "loss": 1.2592,
+      "step": 1465
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.534486888317165e-06,
+      "loss": 1.2013,
+      "step": 1470
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.531365250634378e-06,
+      "loss": 1.2442,
+      "step": 1475
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.52823369588333e-06,
+      "loss": 1.2093,
+      "step": 1480
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.525092230917544e-06,
+      "loss": 1.268,
+      "step": 1485
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.521940862612238e-06,
+      "loss": 1.2355,
+      "step": 1490
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.518779597864308e-06,
+      "loss": 1.2574,
+      "step": 1495
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.515608443592299e-06,
+      "loss": 1.2691,
+      "step": 1500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.512427406736408e-06,
+      "loss": 1.2654,
+      "step": 1505
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.509236494258459e-06,
+      "loss": 1.2545,
+      "step": 1510
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.506035713141883e-06,
+      "loss": 1.2126,
+      "step": 1515
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.502825070391716e-06,
+      "loss": 1.2408,
+      "step": 1520
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.499604573034573e-06,
+      "loss": 1.2606,
+      "step": 1525
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.49637422811864e-06,
+      "loss": 1.2559,
+      "step": 1530
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.493134042713648e-06,
+      "loss": 1.2271,
+      "step": 1535
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.48988402391087e-06,
+      "loss": 1.1991,
+      "step": 1540
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.486624178823096e-06,
+      "loss": 1.2146,
+      "step": 1545
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.483354514584627e-06,
+      "loss": 1.2348,
+      "step": 1550
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.480075038351247e-06,
+      "loss": 1.2658,
+      "step": 1555
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.476785757300218e-06,
+      "loss": 1.2381,
+      "step": 1560
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.473486678630258e-06,
+      "loss": 1.2132,
+      "step": 1565
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.47017780956153e-06,
+      "loss": 1.2107,
+      "step": 1570
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.466859157335616e-06,
+      "loss": 1.2225,
+      "step": 1575
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.463530729215525e-06,
+      "loss": 1.2764,
+      "step": 1580
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.460192532485642e-06,
+      "loss": 1.2234,
+      "step": 1585
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.456844574451744e-06,
+      "loss": 1.2226,
+      "step": 1590
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.453486862440966e-06,
+      "loss": 1.238,
+      "step": 1595
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.450119403801792e-06,
+      "loss": 1.2171,
+      "step": 1600
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.446742205904033e-06,
+      "loss": 1.2226,
+      "step": 1605
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.443355276138818e-06,
+      "loss": 1.2035,
+      "step": 1610
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.439958621918574e-06,
+      "loss": 1.2131,
+      "step": 1615
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.436552250677013e-06,
+      "loss": 1.2377,
+      "step": 1620
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.433136169869106e-06,
+      "loss": 1.2182,
+      "step": 1625
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.42971038697108e-06,
+      "loss": 1.2777,
+      "step": 1630
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.426274909480396e-06,
+      "loss": 1.2405,
+      "step": 1635
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.422829744915724e-06,
+      "loss": 1.2263,
+      "step": 1640
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.419374900816945e-06,
+      "loss": 1.2349,
+      "step": 1645
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.415910384745116e-06,
+      "loss": 1.2043,
+      "step": 1650
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.412436204282466e-06,
+      "loss": 1.1849,
+      "step": 1655
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.408952367032374e-06,
+      "loss": 1.247,
+      "step": 1660
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.405458880619349e-06,
+      "loss": 1.1786,
+      "step": 1665
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.401955752689028e-06,
+      "loss": 1.2198,
+      "step": 1670
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.398442990908134e-06,
+      "loss": 1.2289,
+      "step": 1675
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.39492060296449e-06,
+      "loss": 1.2274,
+      "step": 1680
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.391388596566973e-06,
+      "loss": 1.25,
+      "step": 1685
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.387846979445516e-06,
+      "loss": 1.2243,
+      "step": 1690
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.384295759351087e-06,
+      "loss": 1.1992,
+      "step": 1695
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.38073494405567e-06,
+      "loss": 1.2448,
+      "step": 1700
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.377164541352244e-06,
+      "loss": 1.174,
+      "step": 1705
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.373584559054775e-06,
+      "loss": 1.2098,
+      "step": 1710
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.369995004998194e-06,
+      "loss": 1.2776,
+      "step": 1715
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.366395887038375e-06,
+      "loss": 1.2248,
+      "step": 1720
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.362787213052131e-06,
+      "loss": 1.2531,
+      "step": 1725
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.359168990937185e-06,
+      "loss": 1.1699,
+      "step": 1730
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.355541228612152e-06,
+      "loss": 1.2779,
+      "step": 1735
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.351903934016534e-06,
+      "loss": 1.2744,
+      "step": 1740
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.348257115110689e-06,
+      "loss": 1.2604,
+      "step": 1745
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.344600779875823e-06,
+      "loss": 1.2822,
+      "step": 1750
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.340934936313966e-06,
+      "loss": 1.2412,
+      "step": 1755
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.337259592447958e-06,
+      "loss": 1.2014,
+      "step": 1760
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.333574756321433e-06,
+      "loss": 1.2505,
+      "step": 1765
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.329880435998797e-06,
+      "loss": 1.2217,
+      "step": 1770
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.326176639565213e-06,
+      "loss": 1.1982,
+      "step": 1775
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.322463375126582e-06,
+      "loss": 1.2393,
+      "step": 1780
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.318740650809527e-06,
+      "loss": 1.2119,
+      "step": 1785
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.315008474761377e-06,
+      "loss": 1.2616,
+      "step": 1790
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.31126685515014e-06,
+      "loss": 1.212,
+      "step": 1795
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.3075158001645e-06,
+      "loss": 1.2452,
+      "step": 1800
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.303755318013781e-06,
+      "loss": 1.2247,
+      "step": 1805
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.299985416927944e-06,
+      "loss": 1.2688,
+      "step": 1810
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.296206105157567e-06,
+      "loss": 1.2336,
+      "step": 1815
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.292417390973818e-06,
+      "loss": 1.2903,
+      "step": 1820
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.288619282668447e-06,
+      "loss": 1.1809,
+      "step": 1825
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.284811788553757e-06,
+      "loss": 1.1478,
+      "step": 1830
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.280994916962598e-06,
+      "loss": 1.1931,
+      "step": 1835
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.277168676248343e-06,
+      "loss": 1.2113,
+      "step": 1840
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.273333074784866e-06,
+      "loss": 1.2857,
+      "step": 1845
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.26948812096653e-06,
+      "loss": 1.2701,
+      "step": 1850
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.265633823208164e-06,
+      "loss": 1.2603,
+      "step": 1855
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.261770189945051e-06,
+      "loss": 1.2784,
+      "step": 1860
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.257897229632899e-06,
+      "loss": 1.1626,
+      "step": 1865
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.254014950747832e-06,
+      "loss": 1.1917,
+      "step": 1870
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.250123361786367e-06,
+      "loss": 1.2627,
+      "step": 1875
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.2462224712654e-06,
+      "loss": 1.2233,
+      "step": 1880
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.24231228772218e-06,
+      "loss": 1.2783,
+      "step": 1885
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.23839281971429e-06,
+      "loss": 1.2814,
+      "step": 1890
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.234464075819646e-06,
+      "loss": 1.2279,
+      "step": 1895
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.23052606463645e-06,
+      "loss": 1.1962,
+      "step": 1900
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.226578794783194e-06,
+      "loss": 1.2925,
+      "step": 1905
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.22262227489863e-06,
+      "loss": 1.2339,
+      "step": 1910
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.218656513641755e-06,
+      "loss": 1.2711,
+      "step": 1915
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.214681519691793e-06,
+      "loss": 1.2424,
+      "step": 1920
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.210697301748169e-06,
+      "loss": 1.2055,
+      "step": 1925
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.2067038685305e-06,
+      "loss": 1.2522,
+      "step": 1930
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.20270122877857e-06,
+      "loss": 1.2236,
+      "step": 1935
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.19868939125231e-06,
+      "loss": 1.2118,
+      "step": 1940
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.19466836473178e-06,
+      "loss": 1.2734,
+      "step": 1945
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.190638158017156e-06,
+      "loss": 1.1754,
+      "step": 1950
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.186598779928699e-06,
+      "loss": 1.2419,
+      "step": 1955
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.182550239306744e-06,
+      "loss": 1.1804,
+      "step": 1960
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.17849254501168e-06,
+      "loss": 1.2179,
+      "step": 1965
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.174425705923926e-06,
+      "loss": 1.2009,
+      "step": 1970
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.170349730943919e-06,
+      "loss": 1.2527,
+      "step": 1975
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.16626462899209e-06,
+      "loss": 1.2335,
+      "step": 1980
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.162170409008837e-06,
+      "loss": 1.2779,
+      "step": 1985
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.158067079954524e-06,
+      "loss": 1.2157,
+      "step": 1990
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.153954650809442e-06,
+      "loss": 1.1972,
+      "step": 1995
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.149833130573804e-06,
+      "loss": 1.2105,
+      "step": 2000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.145702528267717e-06,
+      "loss": 1.2661,
+      "step": 2005
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.141562852931161e-06,
+      "loss": 1.2491,
+      "step": 2010
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.137414113623978e-06,
+      "loss": 1.2591,
+      "step": 2015
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.133256319425842e-06,
+      "loss": 1.2602,
+      "step": 2020
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.129089479436249e-06,
+      "loss": 1.205,
+      "step": 2025
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.12491360277449e-06,
+      "loss": 1.2094,
+      "step": 2030
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.120728698579631e-06,
+      "loss": 1.2246,
+      "step": 2035
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.116534776010498e-06,
+      "loss": 1.1941,
+      "step": 2040
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.112331844245652e-06,
+      "loss": 1.2306,
+      "step": 2045
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.108119912483373e-06,
+      "loss": 1.1952,
+      "step": 2050
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.103898989941638e-06,
+      "loss": 1.2694,
+      "step": 2055
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.099669085858094e-06,
+      "loss": 1.2486,
+      "step": 2060
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.095430209490055e-06,
+      "loss": 1.2793,
+      "step": 2065
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.091182370114464e-06,
+      "loss": 1.2487,
+      "step": 2070
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.086925577027883e-06,
+      "loss": 1.2174,
+      "step": 2075
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.082659839546467e-06,
+      "loss": 1.2356,
+      "step": 2080
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.078385167005947e-06,
+      "loss": 1.2258,
+      "step": 2085
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.074101568761614e-06,
+      "loss": 1.1835,
+      "step": 2090
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.069809054188281e-06,
+      "loss": 1.2433,
+      "step": 2095
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.065507632680288e-06,
+      "loss": 1.1928,
+      "step": 2100
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.061197313651461e-06,
+      "loss": 1.2483,
+      "step": 2105
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.056878106535103e-06,
+      "loss": 1.2212,
+      "step": 2110
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.05255002078396e-06,
+      "loss": 1.2793,
+      "step": 2115
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.048213065870218e-06,
+      "loss": 1.2435,
+      "step": 2120
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.043867251285471e-06,
+      "loss": 1.2285,
+      "step": 2125
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.039512586540704e-06,
+      "loss": 1.2551,
+      "step": 2130
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.035149081166266e-06,
+      "loss": 1.2745,
+      "step": 2135
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.030776744711863e-06,
+      "loss": 1.2519,
+      "step": 2140
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.026395586746514e-06,
+      "loss": 1.2055,
+      "step": 2145
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.02200561685856e-06,
+      "loss": 1.2822,
+      "step": 2150
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.017606844655617e-06,
+      "loss": 1.2673,
+      "step": 2155
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.013199279764568e-06,
+      "loss": 1.2026,
+      "step": 2160
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.008782931831542e-06,
+      "loss": 1.2051,
+      "step": 2165
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.004357810521884e-06,
+      "loss": 1.2396,
+      "step": 2170
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.999923925520146e-06,
+      "loss": 1.1841,
+      "step": 2175
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.995481286530057e-06,
+      "loss": 1.2257,
+      "step": 2180
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.991029903274502e-06,
+      "loss": 1.225,
+      "step": 2185
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.986569785495507e-06,
+      "loss": 1.2269,
+      "step": 2190
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.982100942954212e-06,
+      "loss": 1.2354,
+      "step": 2195
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.977623385430853e-06,
+      "loss": 1.2075,
+      "step": 2200
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.973137122724737e-06,
+      "loss": 1.2428,
+      "step": 2205
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.968642164654224e-06,
+      "loss": 1.2487,
+      "step": 2210
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.964138521056702e-06,
+      "loss": 1.2172,
+      "step": 2215
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.959626201788572e-06,
+      "loss": 1.2484,
+      "step": 2220
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.955105216725218e-06,
+      "loss": 1.2855,
+      "step": 2225
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.950575575760992e-06,
+      "loss": 1.1771,
+      "step": 2230
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.946037288809187e-06,
+      "loss": 1.2179,
+      "step": 2235
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.941490365802019e-06,
+      "loss": 1.1735,
+      "step": 2240
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.936934816690608e-06,
+      "loss": 1.2504,
+      "step": 2245
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.932370651444948e-06,
+      "loss": 1.2025,
+      "step": 2250
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.927797880053891e-06,
+      "loss": 1.2622,
+      "step": 2255
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.923216512525125e-06,
+      "loss": 1.1986,
+      "step": 2260
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.918626558885148e-06,
+      "loss": 1.2653,
+      "step": 2265
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.914028029179256e-06,
+      "loss": 1.2053,
+      "step": 2270
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.909420933471504e-06,
+      "loss": 1.2075,
+      "step": 2275
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.904805281844701e-06,
+      "loss": 1.1797,
+      "step": 2280
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.900181084400382e-06,
+      "loss": 1.2713,
+      "step": 2285
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.895548351258777e-06,
+      "loss": 1.1784,
+      "step": 2290
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.890907092558808e-06,
+      "loss": 1.2071,
+      "step": 2295
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.886257318458043e-06,
+      "loss": 1.2129,
+      "step": 2300
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.881599039132696e-06,
+      "loss": 1.2184,
+      "step": 2305
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.876932264777592e-06,
+      "loss": 1.2125,
+      "step": 2310
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.872257005606146e-06,
+      "loss": 1.1682,
+      "step": 2315
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.867573271850345e-06,
+      "loss": 1.2314,
+      "step": 2320
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.862881073760721e-06,
+      "loss": 1.2905,
+      "step": 2325
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.858180421606331e-06,
+      "loss": 1.2555,
+      "step": 2330
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.853471325674734e-06,
+      "loss": 1.2296,
+      "step": 2335
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.848753796271969e-06,
+      "loss": 1.2571,
+      "step": 2340
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.844027843722532e-06,
+      "loss": 1.259,
+      "step": 2345
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.839293478369353e-06,
+      "loss": 1.2189,
+      "step": 2350
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.834550710573773e-06,
+      "loss": 1.26,
+      "step": 2355
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.829799550715523e-06,
+      "loss": 1.2393,
+      "step": 2360
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.825040009192702e-06,
+      "loss": 1.2267,
+      "step": 2365
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.820272096421751e-06,
+      "loss": 1.2461,
+      "step": 2370
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.815495822837428e-06,
+      "loss": 1.1922,
+      "step": 2375
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.810711198892796e-06,
+      "loss": 1.2045,
+      "step": 2380
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.805918235059187e-06,
+      "loss": 1.2488,
+      "step": 2385
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.801116941826193e-06,
+      "loss": 1.2355,
+      "step": 2390
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.796307329701625e-06,
+      "loss": 1.1635,
+      "step": 2395
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.791489409211507e-06,
+      "loss": 1.2351,
+      "step": 2400
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.786663190900044e-06,
+      "loss": 1.2343,
+      "step": 2405
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.781828685329604e-06,
+      "loss": 1.1944,
+      "step": 2410
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.776985903080687e-06,
+      "loss": 1.2083,
+      "step": 2415
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.772134854751911e-06,
+      "loss": 1.1828,
+      "step": 2420
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.76727555095998e-06,
+      "loss": 1.188,
+      "step": 2425
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.762408002339672e-06,
+      "loss": 1.2454,
+      "step": 2430
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.757532219543802e-06,
+      "loss": 1.2282,
+      "step": 2435
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.752648213243215e-06,
+      "loss": 1.2582,
+      "step": 2440
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.747755994126744e-06,
+      "loss": 1.1919,
+      "step": 2445
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.7428555729012e-06,
+      "loss": 1.1555,
+      "step": 2450
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.737946960291345e-06,
+      "loss": 1.2121,
+      "step": 2455
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.733030167039867e-06,
+      "loss": 1.1924,
+      "step": 2460
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.72810520390736e-06,
+      "loss": 1.2168,
+      "step": 2465
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.723172081672296e-06,
+      "loss": 1.2029,
+      "step": 2470
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.718230811131001e-06,
+      "loss": 1.2274,
+      "step": 2475
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.713281403097641e-06,
+      "loss": 1.2323,
+      "step": 2480
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.708323868404185e-06,
+      "loss": 1.2312,
+      "step": 2485
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.703358217900393e-06,
+      "loss": 1.2186,
+      "step": 2490
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.698384462453778e-06,
+      "loss": 1.2791,
+      "step": 2495
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.693402612949598e-06,
+      "loss": 1.2184,
+      "step": 2500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.688412680290824e-06,
+      "loss": 1.1967,
+      "step": 2505
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.683414675398118e-06,
+      "loss": 1.2046,
+      "step": 2510
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.678408609209808e-06,
+      "loss": 1.2623,
+      "step": 2515
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.67339449268186e-06,
+      "loss": 1.282,
+      "step": 2520
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.668372336787862e-06,
+      "loss": 1.2712,
+      "step": 2525
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.663342152519001e-06,
+      "loss": 1.2462,
+      "step": 2530
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.658303950884028e-06,
+      "loss": 1.2301,
+      "step": 2535
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.65325774290924e-06,
+      "loss": 1.2617,
+      "step": 2540
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.648203539638463e-06,
+      "loss": 1.226,
+      "step": 2545
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.643141352133012e-06,
+      "loss": 1.266,
+      "step": 2550
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.638071191471683e-06,
+      "loss": 1.1974,
+      "step": 2555
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.632993068750716e-06,
+      "loss": 1.2307,
+      "step": 2560
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.627906995083783e-06,
+      "loss": 1.2828,
+      "step": 2565
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.622812981601952e-06,
+      "loss": 1.2055,
+      "step": 2570
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.617711039453667e-06,
+      "loss": 1.1971,
+      "step": 2575
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.612601179804726e-06,
+      "loss": 1.2713,
+      "step": 2580
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.607483413838257e-06,
+      "loss": 1.2934,
+      "step": 2585
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.602357752754687e-06,
+      "loss": 1.2552,
+      "step": 2590
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.597224207771724e-06,
+      "loss": 1.1734,
+      "step": 2595
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.59208279012433e-06,
+      "loss": 1.222,
+      "step": 2600
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.586933511064697e-06,
+      "loss": 1.2552,
+      "step": 2605
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.581776381862221e-06,
+      "loss": 1.2447,
+      "step": 2610
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.576611413803481e-06,
+      "loss": 1.2373,
+      "step": 2615
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.571438618192205e-06,
+      "loss": 1.2734,
+      "step": 2620
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.56625800634926e-06,
+      "loss": 1.2762,
+      "step": 2625
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.561069589612616e-06,
+      "loss": 1.2284,
+      "step": 2630
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.55587337933732e-06,
+      "loss": 1.2375,
+      "step": 2635
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.550669386895485e-06,
+      "loss": 1.2465,
+      "step": 2640
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.545457623676245e-06,
+      "loss": 1.2921,
+      "step": 2645
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.540238101085748e-06,
+      "loss": 1.274,
+      "step": 2650
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.535010830547119e-06,
+      "loss": 1.1872,
+      "step": 2655
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.529775823500445e-06,
+      "loss": 1.2393,
+      "step": 2660
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.52453309140274e-06,
+      "loss": 1.2729,
+      "step": 2665
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.519282645727926e-06,
+      "loss": 1.2459,
+      "step": 2670
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.514024497966806e-06,
+      "loss": 1.2547,
+      "step": 2675
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.508758659627044e-06,
+      "loss": 1.2458,
+      "step": 2680
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.503485142233123e-06,
+      "loss": 1.1926,
+      "step": 2685
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.498203957326348e-06,
+      "loss": 1.2358,
+      "step": 2690
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.492915116464793e-06,
+      "loss": 1.2535,
+      "step": 2695
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.487618631223292e-06,
+      "loss": 1.2301,
+      "step": 2700
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.482314513193405e-06,
+      "loss": 1.2693,
+      "step": 2705
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.477002773983402e-06,
+      "loss": 1.2129,
+      "step": 2710
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.471683425218233e-06,
+      "loss": 1.1915,
+      "step": 2715
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.466356478539492e-06,
+      "loss": 1.2216,
+      "step": 2720
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.461021945605414e-06,
+      "loss": 1.2107,
+      "step": 2725
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.45567983809083e-06,
+      "loss": 1.2027,
+      "step": 2730
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.450330167687145e-06,
+      "loss": 1.24,
+      "step": 2735
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.444972946102322e-06,
+      "loss": 1.2375,
+      "step": 2740
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.439608185060847e-06,
+      "loss": 1.2444,
+      "step": 2745
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.43423589630371e-06,
+      "loss": 1.223,
+      "step": 2750
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.428856091588371e-06,
+      "loss": 1.2897,
+      "step": 2755
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.423468782688742e-06,
+      "loss": 1.2659,
+      "step": 2760
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.418073981395154e-06,
+      "loss": 1.2025,
+      "step": 2765
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.412671699514344e-06,
+      "loss": 1.2103,
+      "step": 2770
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.407261948869408e-06,
+      "loss": 1.2428,
+      "step": 2775
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.4018447412998e-06,
+      "loss": 1.2459,
+      "step": 2780
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.396420088661288e-06,
+      "loss": 1.2237,
+      "step": 2785
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.390988002825931e-06,
+      "loss": 1.2496,
+      "step": 2790
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.385548495682064e-06,
+      "loss": 1.1963,
+      "step": 2795
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.380101579134253e-06,
+      "loss": 1.2577,
+      "step": 2800
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.37464726510329e-06,
+      "loss": 1.1994,
+      "step": 2805
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.369185565526152e-06,
+      "loss": 1.233,
+      "step": 2810
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.36371649235598e-06,
+      "loss": 1.2643,
+      "step": 2815
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.35824005756205e-06,
+      "loss": 1.2162,
+      "step": 2820
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.352756273129754e-06,
+      "loss": 1.221,
+      "step": 2825
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.347265151060565e-06,
+      "loss": 1.2717,
+      "step": 2830
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.341766703372018e-06,
+      "loss": 1.1876,
+      "step": 2835
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.336260942097677e-06,
+      "loss": 1.2074,
+      "step": 2840
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.330747879287113e-06,
+      "loss": 1.241,
+      "step": 2845
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.325227527005882e-06,
+      "loss": 1.1824,
+      "step": 2850
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.319699897335481e-06,
+      "loss": 1.1995,
+      "step": 2855
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.314165002373349e-06,
+      "loss": 1.2089,
+      "step": 2860
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.308622854232813e-06,
+      "loss": 1.2289,
+      "step": 2865
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.303073465043082e-06,
+      "loss": 1.2583,
+      "step": 2870
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.297516846949205e-06,
+      "loss": 1.2434,
+      "step": 2875
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.291953012112059e-06,
+      "loss": 1.237,
+      "step": 2880
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.286381972708312e-06,
+      "loss": 1.1752,
+      "step": 2885
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.280803740930397e-06,
+      "loss": 1.2871,
+      "step": 2890
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.275218328986495e-06,
+      "loss": 1.2307,
+      "step": 2895
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.26962574910049e-06,
+      "loss": 1.2672,
+      "step": 2900
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.264026013511963e-06,
+      "loss": 1.258,
+      "step": 2905
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.258419134476153e-06,
+      "loss": 1.1937,
+      "step": 2910
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.25280512426393e-06,
+      "loss": 1.2247,
+      "step": 2915
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.247183995161774e-06,
+      "loss": 1.2339,
+      "step": 2920
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.24155575947174e-06,
+      "loss": 1.1705,
+      "step": 2925
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.235920429511446e-06,
+      "loss": 1.2424,
+      "step": 2930
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.230278017614024e-06,
+      "loss": 1.2242,
+      "step": 2935
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.224628536128114e-06,
+      "loss": 1.1931,
+      "step": 2940
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.218971997417824e-06,
+      "loss": 1.2189,
+      "step": 2945
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.213308413862707e-06,
+      "loss": 1.2524,
+      "step": 2950
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.207637797857737e-06,
+      "loss": 1.2658,
+      "step": 2955
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.201960161813276e-06,
+      "loss": 1.2396,
+      "step": 2960
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.196275518155048e-06,
+      "loss": 1.2209,
+      "step": 2965
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.19058387932412e-06,
+      "loss": 1.1996,
+      "step": 2970
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.18488525777686e-06,
+      "loss": 1.2233,
+      "step": 2975
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.179179665984922e-06,
+      "loss": 1.1898,
+      "step": 2980
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.173467116435218e-06,
+      "loss": 1.2393,
+      "step": 2985
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.16774762162988e-06,
+      "loss": 1.2518,
+      "step": 2990
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.162021194086246e-06,
+      "loss": 1.2343,
+      "step": 2995
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.156287846336823e-06,
+      "loss": 1.2181,
+      "step": 3000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.150547590929267e-06,
+      "loss": 1.2471,
+      "step": 3005
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.144800440426345e-06,
+      "loss": 1.2433,
+      "step": 3010
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.139046407405921e-06,
+      "loss": 1.2464,
+      "step": 3015
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.13328550446092e-06,
+      "loss": 1.1921,
+      "step": 3020
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.1275177441993e-06,
+      "loss": 1.1828,
+      "step": 3025
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.121743139244027e-06,
+      "loss": 1.2437,
+      "step": 3030
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.115961702233048e-06,
+      "loss": 1.2355,
+      "step": 3035
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.110173445819263e-06,
+      "loss": 1.2269,
+      "step": 3040
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.104378382670491e-06,
+      "loss": 1.1938,
+      "step": 3045
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.098576525469457e-06,
+      "loss": 1.2,
+      "step": 3050
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.092767886913748e-06,
+      "loss": 1.2103,
+      "step": 3055
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.086952479715793e-06,
+      "loss": 1.2651,
+      "step": 3060
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.081130316602838e-06,
+      "loss": 1.1872,
+      "step": 3065
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.075301410316912e-06,
+      "loss": 1.255,
+      "step": 3070
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.0694657736148e-06,
+      "loss": 1.2627,
+      "step": 3075
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.06362341926802e-06,
+      "loss": 1.1728,
+      "step": 3080
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.057774360062793e-06,
+      "loss": 1.2404,
+      "step": 3085
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.051918608800009e-06,
+      "loss": 1.2136,
+      "step": 3090
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.046056178295206e-06,
+      "loss": 1.2057,
+      "step": 3095
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.040187081378542e-06,
+      "loss": 1.2423,
+      "step": 3100
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.034311330894761e-06,
+      "loss": 1.1803,
+      "step": 3105
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.02842893970317e-06,
+      "loss": 1.2006,
+      "step": 3110
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.022539920677613e-06,
+      "loss": 1.1922,
+      "step": 3115
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.016644286706433e-06,
+      "loss": 1.2263,
+      "step": 3120
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.010742050692455e-06,
+      "loss": 1.2477,
+      "step": 3125
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.00483322555295e-06,
+      "loss": 1.226,
+      "step": 3130
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.99891782421961e-06,
+      "loss": 1.265,
+      "step": 3135
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.99299585963852e-06,
+      "loss": 1.227,
+      "step": 3140
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.98706734477013e-06,
+      "loss": 1.2216,
+      "step": 3145
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.981132292589224e-06,
+      "loss": 1.2616,
+      "step": 3150
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.975190716084893e-06,
+      "loss": 1.1903,
+      "step": 3155
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.969242628260507e-06,
+      "loss": 1.2308,
+      "step": 3160
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.963288042133686e-06,
+      "loss": 1.3016,
+      "step": 3165
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.957326970736275e-06,
+      "loss": 1.1936,
+      "step": 3170
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.951359427114306e-06,
+      "loss": 1.2297,
+      "step": 3175
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.945385424327981e-06,
+      "loss": 1.2291,
+      "step": 3180
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.939404975451636e-06,
+      "loss": 1.2212,
+      "step": 3185
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.933418093573714e-06,
+      "loss": 1.2414,
+      "step": 3190
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.927424791796738e-06,
+      "loss": 1.2382,
+      "step": 3195
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.921425083237284e-06,
+      "loss": 1.2401,
+      "step": 3200
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.915418981025943e-06,
+      "loss": 1.186,
+      "step": 3205
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.909406498307303e-06,
+      "loss": 1.1709,
+      "step": 3210
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.903387648239916e-06,
+      "loss": 1.2169,
+      "step": 3215
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.897362443996268e-06,
+      "loss": 1.2647,
+      "step": 3220
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.89133089876275e-06,
+      "loss": 1.2416,
+      "step": 3225
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.885293025739634e-06,
+      "loss": 1.1713,
+      "step": 3230
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.87924883814104e-06,
+      "loss": 1.2494,
+      "step": 3235
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.873198349194903e-06,
+      "loss": 1.1865,
+      "step": 3240
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.867141572142958e-06,
+      "loss": 1.2434,
+      "step": 3245
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.861078520240689e-06,
+      "loss": 1.2221,
+      "step": 3250
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.855009206757325e-06,
+      "loss": 1.2389,
+      "step": 3255
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.84893364497579e-06,
+      "loss": 1.2234,
+      "step": 3260
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.842851848192688e-06,
+      "loss": 1.2358,
+      "step": 3265
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.836763829718269e-06,
+      "loss": 1.2015,
+      "step": 3270
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.830669602876393e-06,
+      "loss": 1.2222,
+      "step": 3275
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.824569181004513e-06,
+      "loss": 1.2535,
+      "step": 3280
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.818462577453639e-06,
+      "loss": 1.2217,
+      "step": 3285
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.812349805588308e-06,
+      "loss": 1.2082,
+      "step": 3290
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.806230878786561e-06,
+      "loss": 1.2427,
+      "step": 3295
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.8001058104399e-06,
+      "loss": 1.2168,
+      "step": 3300
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.793974613953281e-06,
+      "loss": 1.1697,
+      "step": 3305
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.78783730274506e-06,
+      "loss": 1.2049,
+      "step": 3310
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.781693890246982e-06,
+      "loss": 1.2151,
+      "step": 3315
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.775544389904142e-06,
+      "loss": 1.2278,
+      "step": 3320
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.769388815174963e-06,
+      "loss": 1.1966,
+      "step": 3325
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.763227179531155e-06,
+      "loss": 1.2076,
+      "step": 3330
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.757059496457698e-06,
+      "loss": 1.2531,
+      "step": 3335
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.750885779452804e-06,
+      "loss": 1.2001,
+      "step": 3340
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.744706042027891e-06,
+      "loss": 1.2402,
+      "step": 3345
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.738520297707556e-06,
+      "loss": 1.2372,
+      "step": 3350
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.73232856002954e-06,
+      "loss": 1.2054,
+      "step": 3355
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.726130842544697e-06,
+      "loss": 1.2171,
+      "step": 3360
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.719927158816976e-06,
+      "loss": 1.2503,
+      "step": 3365
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.713717522423374e-06,
+      "loss": 1.2144,
+      "step": 3370
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.707501946953926e-06,
+      "loss": 1.1641,
+      "step": 3375
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.701280446011653e-06,
+      "loss": 1.2765,
+      "step": 3380
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.695053033212558e-06,
+      "loss": 1.261,
+      "step": 3385
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.688819722185572e-06,
+      "loss": 1.2039,
+      "step": 3390
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.682580526572535e-06,
+      "loss": 1.1917,
+      "step": 3395
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.67633546002817e-06,
+      "loss": 1.1857,
+      "step": 3400
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.670084536220051e-06,
+      "loss": 1.2213,
+      "step": 3405
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.663827768828558e-06,
+      "loss": 1.187,
+      "step": 3410
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.657565171546878e-06,
+      "loss": 1.1954,
+      "step": 3415
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.65129675808094e-06,
+      "loss": 1.2657,
+      "step": 3420
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.645022542149412e-06,
+      "loss": 1.2075,
+      "step": 3425
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.63874253748366e-06,
+      "loss": 1.1668,
+      "step": 3430
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.632456757827718e-06,
+      "loss": 1.2017,
+      "step": 3435
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.626165216938255e-06,
+      "loss": 1.1722,
+      "step": 3440
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.619867928584552e-06,
+      "loss": 1.2087,
+      "step": 3445
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.61356490654847e-06,
+      "loss": 1.2347,
+      "step": 3450
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.607256164624418e-06,
+      "loss": 1.189,
+      "step": 3455
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.600941716619319e-06,
+      "loss": 1.2171,
+      "step": 3460
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.594621576352589e-06,
+      "loss": 1.2067,
+      "step": 3465
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.588295757656098e-06,
+      "loss": 1.1861,
+      "step": 3470
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.581964274374145e-06,
+      "loss": 1.2247,
+      "step": 3475
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.575627140363429e-06,
+      "loss": 1.2087,
+      "step": 3480
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.5692843694930105e-06,
+      "loss": 1.2169,
+      "step": 3485
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.56293597564429e-06,
+      "loss": 1.213,
+      "step": 3490
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.556581972710972e-06,
+      "loss": 1.2218,
+      "step": 3495
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.550222374599039e-06,
+      "loss": 1.2029,
+      "step": 3500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.543857195226718e-06,
+      "loss": 1.2768,
+      "step": 3505
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.537486448524449e-06,
+      "loss": 1.1583,
+      "step": 3510
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.5311101484348605e-06,
+      "loss": 1.2053,
+      "step": 3515
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.52472830891273e-06,
+      "loss": 1.2213,
+      "step": 3520
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.51834094392496e-06,
+      "loss": 1.2203,
+      "step": 3525
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.511948067450547e-06,
+      "loss": 1.2004,
+      "step": 3530
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.505549693480548e-06,
+      "loss": 1.2027,
+      "step": 3535
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.499145836018053e-06,
+      "loss": 1.2387,
+      "step": 3540
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.492736509078151e-06,
+      "loss": 1.2813,
+      "step": 3545
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.486321726687902e-06,
+      "loss": 1.2308,
+      "step": 3550
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.479901502886309e-06,
+      "loss": 1.2244,
+      "step": 3555
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.4734758517242744e-06,
+      "loss": 1.192,
+      "step": 3560
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.467044787264588e-06,
+      "loss": 1.1914,
+      "step": 3565
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.460608323581883e-06,
+      "loss": 1.2356,
+      "step": 3570
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.45416647476261e-06,
+      "loss": 1.196,
+      "step": 3575
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.447719254905003e-06,
+      "loss": 1.2119,
+      "step": 3580
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.441266678119052e-06,
+      "loss": 1.198,
+      "step": 3585
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.4348087585264735e-06,
+      "loss": 1.1974,
+      "step": 3590
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.4283455102606725e-06,
+      "loss": 1.1709,
+      "step": 3595
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.421876947466721e-06,
+      "loss": 1.2211,
+      "step": 3600
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.415403084301313e-06,
+      "loss": 1.2434,
+      "step": 3605
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.4089239349327545e-06,
+      "loss": 1.2492,
+      "step": 3610
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.402439513540912e-06,
+      "loss": 1.2323,
+      "step": 3615
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.3959498343171934e-06,
+      "loss": 1.2079,
+      "step": 3620
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.3894549114645135e-06,
+      "loss": 1.1491,
+      "step": 3625
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.38295475919726e-06,
+      "loss": 1.2059,
+      "step": 3630
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.376449391741269e-06,
+      "loss": 1.2593,
+      "step": 3635
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.3699388233337905e-06,
+      "loss": 1.2003,
+      "step": 3640
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.3634230682234565e-06,
+      "loss": 1.2113,
+      "step": 3645
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.356902140670245e-06,
+      "loss": 1.2085,
+      "step": 3650
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.3503760549454604e-06,
+      "loss": 1.1613,
+      "step": 3655
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.343844825331694e-06,
+      "loss": 1.2167,
+      "step": 3660
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.337308466122797e-06,
+      "loss": 1.2306,
+      "step": 3665
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.33076699162384e-06,
+      "loss": 1.2402,
+      "step": 3670
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.324220416151097e-06,
+      "loss": 1.2608,
+      "step": 3675
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.317668754031999e-06,
+      "loss": 1.1872,
+      "step": 3680
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.311112019605113e-06,
+      "loss": 1.2006,
+      "step": 3685
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.304550227220109e-06,
+      "loss": 1.2038,
+      "step": 3690
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.29798339123772e-06,
+      "loss": 1.2772,
+      "step": 3695
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.291411526029721e-06,
+      "loss": 1.238,
+      "step": 3700
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.284834645978895e-06,
+      "loss": 1.2405,
+      "step": 3705
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.278252765479e-06,
+      "loss": 1.2214,
+      "step": 3710
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.271665898934731e-06,
+      "loss": 1.1704,
+      "step": 3715
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.265074060761707e-06,
+      "loss": 1.1888,
+      "step": 3720
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.258477265386416e-06,
+      "loss": 1.244,
+      "step": 3725
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.251875527246202e-06,
+      "loss": 1.2039,
+      "step": 3730
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.245268860789224e-06,
+      "loss": 1.1615,
+      "step": 3735
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.238657280474427e-06,
+      "loss": 1.2295,
+      "step": 3740
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.2320408007715125e-06,
+      "loss": 1.2395,
+      "step": 3745
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.225419436160901e-06,
+      "loss": 1.2437,
+      "step": 3750
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.218793201133703e-06,
+      "loss": 1.197,
+      "step": 3755
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.212162110191694e-06,
+      "loss": 1.203,
+      "step": 3760
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.205526177847273e-06,
+      "loss": 1.1997,
+      "step": 3765
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.198885418623432e-06,
+      "loss": 1.2124,
+      "step": 3770
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.192239847053732e-06,
+      "loss": 1.2156,
+      "step": 3775
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.185589477682262e-06,
+      "loss": 1.2385,
+      "step": 3780
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.178934325063615e-06,
+      "loss": 1.2015,
+      "step": 3785
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.1722744037628475e-06,
+      "loss": 1.2082,
+      "step": 3790
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.165609728355458e-06,
+      "loss": 1.2339,
+      "step": 3795
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.158940313427345e-06,
+      "loss": 1.2596,
+      "step": 3800
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.15226617357478e-06,
+      "loss": 1.2164,
+      "step": 3805
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.145587323404379e-06,
+      "loss": 1.2783,
+      "step": 3810
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.1389037775330615e-06,
+      "loss": 1.2137,
+      "step": 3815
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.132215550588029e-06,
+      "loss": 1.2264,
+      "step": 3820
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.125522657206723e-06,
+      "loss": 1.1773,
+      "step": 3825
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.1188251120368e-06,
+      "loss": 1.2011,
+      "step": 3830
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.112122929736099e-06,
+      "loss": 1.2112,
+      "step": 3835
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.105416124972604e-06,
+      "loss": 1.2051,
+      "step": 3840
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.098704712424416e-06,
+      "loss": 1.2367,
+      "step": 3845
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.091988706779722e-06,
+      "loss": 1.2704,
+      "step": 3850
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.085268122736759e-06,
+      "loss": 1.2379,
+      "step": 3855
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.0785429750037886e-06,
+      "loss": 1.1949,
+      "step": 3860
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.071813278299054e-06,
+      "loss": 1.223,
+      "step": 3865
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.065079047350758e-06,
+      "loss": 1.2187,
+      "step": 3870
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.058340296897026e-06,
+      "loss": 1.2694,
+      "step": 3875
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.0515970416858715e-06,
+      "loss": 1.2184,
+      "step": 3880
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.0448492964751744e-06,
+      "loss": 1.2705,
+      "step": 3885
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.038097076032634e-06,
+      "loss": 1.1638,
+      "step": 3890
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.031340395135744e-06,
+      "loss": 1.19,
+      "step": 3895
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.024579268571765e-06,
+      "loss": 1.1315,
+      "step": 3900
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.017813711137683e-06,
+      "loss": 1.2275,
+      "step": 3905
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.011043737640183e-06,
+      "loss": 1.1985,
+      "step": 3910
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.004269362895615e-06,
+      "loss": 1.2149,
+      "step": 3915
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.997490601729957e-06,
+      "loss": 1.1797,
+      "step": 3920
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.9907074689787926e-06,
+      "loss": 1.2348,
+      "step": 3925
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.98391997948727e-06,
+      "loss": 1.2054,
+      "step": 3930
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.9771281481100725e-06,
+      "loss": 1.1544,
+      "step": 3935
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.9703319897113864e-06,
+      "loss": 1.2191,
+      "step": 3940
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.963531519164868e-06,
+      "loss": 1.1916,
+      "step": 3945
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.956726751353608e-06,
+      "loss": 1.1639,
+      "step": 3950
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.949917701170108e-06,
+      "loss": 1.2133,
+      "step": 3955
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.943104383516235e-06,
+      "loss": 1.2159,
+      "step": 3960
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.936286813303199e-06,
+      "loss": 1.1678,
+      "step": 3965
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.9294650054515155e-06,
+      "loss": 1.1663,
+      "step": 3970
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.922638974890978e-06,
+      "loss": 1.1957,
+      "step": 3975
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.915808736560614e-06,
+      "loss": 1.1947,
+      "step": 3980
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.90897430540867e-06,
+      "loss": 1.2102,
+      "step": 3985
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.902135696392559e-06,
+      "loss": 1.2129,
+      "step": 3990
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.895292924478842e-06,
+      "loss": 1.2128,
+      "step": 3995
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.888446004643192e-06,
+      "loss": 1.1944,
+      "step": 4000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.881594951870357e-06,
+      "loss": 1.1894,
+      "step": 4005
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.874739781154132e-06,
+      "loss": 1.2022,
+      "step": 4010
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.867880507497322e-06,
+      "loss": 1.2787,
+      "step": 4015
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.861017145911715e-06,
+      "loss": 1.1975,
+      "step": 4020
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.854149711418041e-06,
+      "loss": 1.2539,
+      "step": 4025
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.847278219045948e-06,
+      "loss": 1.2487,
+      "step": 4030
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.8404026838339635e-06,
+      "loss": 1.2308,
+      "step": 4035
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.83352312082946e-06,
+      "loss": 1.2168,
+      "step": 4040
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.82663954508863e-06,
+      "loss": 1.1773,
+      "step": 4045
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.819751971676445e-06,
+      "loss": 1.2334,
+      "step": 4050
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.812860415666625e-06,
+      "loss": 1.1629,
+      "step": 4055
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.805964892141608e-06,
+      "loss": 1.2163,
+      "step": 4060
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.799065416192512e-06,
+      "loss": 1.2282,
+      "step": 4065
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.79216200291911e-06,
+      "loss": 1.1721,
+      "step": 4070
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.785254667429789e-06,
+      "loss": 1.1628,
+      "step": 4075
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.77834342484152e-06,
+      "loss": 1.2204,
+      "step": 4080
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.771428290279823e-06,
+      "loss": 1.1416,
+      "step": 4085
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.764509278878739e-06,
+      "loss": 1.1805,
+      "step": 4090
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.7575864057807925e-06,
+      "loss": 1.2613,
+      "step": 4095
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.7506596861369585e-06,
+      "loss": 1.2193,
+      "step": 4100
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.743729135106634e-06,
+      "loss": 1.1889,
+      "step": 4105
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.736794767857593e-06,
+      "loss": 1.249,
+      "step": 4110
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.72985659956597e-06,
+      "loss": 1.2622,
+      "step": 4115
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.722914645416215e-06,
+      "loss": 1.2191,
+      "step": 4120
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.715968920601063e-06,
+      "loss": 1.1936,
+      "step": 4125
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.7090194403215005e-06,
+      "loss": 1.3156,
+      "step": 4130
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.7020662197867345e-06,
+      "loss": 1.214,
+      "step": 4135
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.695109274214159e-06,
+      "loss": 1.2034,
+      "step": 4140
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.688148618829317e-06,
+      "loss": 1.2401,
+      "step": 4145
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.681184268865872e-06,
+      "loss": 1.1809,
+      "step": 4150
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.674216239565576e-06,
+      "loss": 1.2267,
+      "step": 4155
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.667244546178228e-06,
+      "loss": 1.2277,
+      "step": 4160
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.6602692039616515e-06,
+      "loss": 1.2366,
+      "step": 4165
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.6532902281816505e-06,
+      "loss": 1.1875,
+      "step": 4170
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.6463076341119834e-06,
+      "loss": 1.1938,
+      "step": 4175
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.639321437034331e-06,
+      "loss": 1.2021,
+      "step": 4180
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.632331652238252e-06,
+      "loss": 1.1998,
+      "step": 4185
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.625338295021162e-06,
+      "loss": 1.2098,
+      "step": 4190
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.618341380688297e-06,
+      "loss": 1.1834,
+      "step": 4195
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.611340924552672e-06,
+      "loss": 1.2557,
+      "step": 4200
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.604336941935058e-06,
+      "loss": 1.1757,
+      "step": 4205
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.5973294481639394e-06,
+      "loss": 1.1653,
+      "step": 4210
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.5903184585754895e-06,
+      "loss": 1.2064,
+      "step": 4215
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.5833039885135304e-06,
+      "loss": 1.2517,
+      "step": 4220
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.576286053329504e-06,
+      "loss": 1.2282,
+      "step": 4225
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.569264668382427e-06,
+      "loss": 1.2097,
+      "step": 4230
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.562239849038876e-06,
+      "loss": 1.2141,
+      "step": 4235
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.555211610672939e-06,
+      "loss": 1.1976,
+      "step": 4240
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.5481799686661885e-06,
+      "loss": 1.1911,
+      "step": 4245
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.5411449384076466e-06,
+      "loss": 1.2159,
+      "step": 4250
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.534106535293745e-06,
+      "loss": 1.2672,
+      "step": 4255
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.527064774728307e-06,
+      "loss": 1.217,
+      "step": 4260
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.520019672122493e-06,
+      "loss": 1.2274,
+      "step": 4265
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.512971242894786e-06,
+      "loss": 1.1312,
+      "step": 4270
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 6.505919502470946e-06,
+      "loss": 1.2732,
+      "step": 4275
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 6.498864466283978e-06,
+      "loss": 1.247,
+      "step": 4280
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 6.491806149774104e-06,
+      "loss": 1.1909,
+      "step": 4285
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 6.484744568388721e-06,
+      "loss": 1.1798,
+      "step": 4290
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 6.4776797375823745e-06,
+      "loss": 1.204,
+      "step": 4295
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 6.47061167281672e-06,
+      "loss": 1.2188,
+      "step": 4300
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 6.46354038956049e-06,
+      "loss": 1.2311,
+      "step": 4305
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 6.456465903289463e-06,
+      "loss": 1.2286,
+      "step": 4310
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 6.449388229486424e-06,
+      "loss": 1.2317,
+      "step": 4315
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 6.44230738364114e-06,
+      "loss": 1.2469,
+      "step": 4320
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 6.435223381250312e-06,
+      "loss": 1.1677,
+      "step": 4325
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 6.428136237817555e-06,
+      "loss": 1.2235,
+      "step": 4330
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 6.421045968853358e-06,
+      "loss": 1.1933,
+      "step": 4335
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 6.4139525898750475e-06,
+      "loss": 1.2784,
+      "step": 4340
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 6.406856116406759e-06,
+      "loss": 1.2077,
+      "step": 4345
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 6.399756563979399e-06,
+      "loss": 1.233,
+      "step": 4350
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 6.392653948130612e-06,
+      "loss": 1.2218,
+      "step": 4355
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 6.385548284404749e-06,
+      "loss": 1.2521,
+      "step": 4360
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 6.37843958835283e-06,
+      "loss": 1.2754,
+      "step": 4365
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 6.371327875532509e-06,
+      "loss": 1.2187,
+      "step": 4370
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 6.3642131615080484e-06,
+      "loss": 1.1939,
+      "step": 4375
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 6.357095461850274e-06,
+      "loss": 1.17,
+      "step": 4380
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.349974792136546e-06,
+      "loss": 1.2572,
+      "step": 4385
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.342851167950728e-06,
+      "loss": 1.2022,
+      "step": 4390
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.335724604883145e-06,
+      "loss": 1.2181,
+      "step": 4395
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.328595118530557e-06,
+      "loss": 1.1926,
+      "step": 4400
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.321462724496121e-06,
+      "loss": 1.1626,
+      "step": 4405
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.314327438389357e-06,
+      "loss": 1.2403,
+      "step": 4410
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.307189275826115e-06,
+      "loss": 1.1694,
+      "step": 4415
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.300048252428541e-06,
+      "loss": 1.2631,
+      "step": 4420
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.29290438382504e-06,
+      "loss": 1.2437,
+      "step": 4425
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.285757685650244e-06,
+      "loss": 1.2903,
+      "step": 4430
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.278608173544981e-06,
+      "loss": 1.1827,
+      "step": 4435
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.271455863156235e-06,
+      "loss": 1.1708,
+      "step": 4440
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.264300770137113e-06,
+      "loss": 1.2226,
+      "step": 4445
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.257142910146812e-06,
+      "loss": 1.1801,
+      "step": 4450
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.249982298850587e-06,
+      "loss": 1.1773,
+      "step": 4455
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.242818951919714e-06,
+      "loss": 1.2259,
+      "step": 4460
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.235652885031453e-06,
+      "loss": 1.2121,
+      "step": 4465
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.22848411386902e-06,
+      "loss": 1.1843,
+      "step": 4470
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.221312654121549e-06,
+      "loss": 1.1318,
+      "step": 4475
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.2141385214840546e-06,
+      "loss": 1.2114,
+      "step": 4480
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.206961731657406e-06,
+      "loss": 1.2366,
+      "step": 4485
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.199782300348285e-06,
+      "loss": 1.17,
+      "step": 4490
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.192600243269156e-06,
+      "loss": 1.2296,
+      "step": 4495
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.1854155761382255e-06,
+      "loss": 1.2345,
+      "step": 4500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.178228314679417e-06,
+      "loss": 1.2075,
+      "step": 4505
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.171038474622335e-06,
+      "loss": 1.2097,
+      "step": 4510
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.163846071702218e-06,
+      "loss": 1.1675,
+      "step": 4515
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.15665112165992e-06,
+      "loss": 1.2223,
+      "step": 4520
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.149453640241866e-06,
+      "loss": 1.1885,
+      "step": 4525
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.1422536432000236e-06,
+      "loss": 1.2363,
+      "step": 4530
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.135051146291866e-06,
+      "loss": 1.1956,
+      "step": 4535
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.127846165280337e-06,
+      "loss": 1.1864,
+      "step": 4540
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.120638715933813e-06,
+      "loss": 1.204,
+      "step": 4545
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.113428814026079e-06,
+      "loss": 1.2122,
+      "step": 4550
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.106216475336284e-06,
+      "loss": 1.1645,
+      "step": 4555
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.099001715648909e-06,
+      "loss": 1.2178,
+      "step": 4560
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.091784550753736e-06,
+      "loss": 1.1878,
+      "step": 4565
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.08456499644581e-06,
+      "loss": 1.2439,
+      "step": 4570
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.0773430685254046e-06,
+      "loss": 1.1814,
+      "step": 4575
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.070118782797989e-06,
+      "loss": 1.1224,
+      "step": 4580
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.062892155074191e-06,
+      "loss": 1.1976,
+      "step": 4585
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.055663201169768e-06,
+      "loss": 1.2058,
+      "step": 4590
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.0484319369055614e-06,
+      "loss": 1.1937,
+      "step": 4595
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.0411983781074755e-06,
+      "loss": 1.2392,
+      "step": 4600
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.033962540606434e-06,
+      "loss": 1.2675,
+      "step": 4605
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.026724440238345e-06,
+      "loss": 1.1912,
+      "step": 4610
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.019484092844076e-06,
+      "loss": 1.2325,
+      "step": 4615
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.0122415142694025e-06,
+      "loss": 1.1694,
+      "step": 4620
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.004996720364988e-06,
+      "loss": 1.2358,
+      "step": 4625
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.997749726986349e-06,
+      "loss": 1.18,
+      "step": 4630
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.990500549993809e-06,
+      "loss": 1.1803,
+      "step": 4635
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.983249205252473e-06,
+      "loss": 1.2042,
+      "step": 4640
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.9759957086321866e-06,
+      "loss": 1.222,
+      "step": 4645
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.968740076007511e-06,
+      "loss": 1.2193,
+      "step": 4650
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.961482323257677e-06,
+      "loss": 1.2022,
+      "step": 4655
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.95422246626656e-06,
+      "loss": 1.2196,
+      "step": 4660
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.9469605209226365e-06,
+      "loss": 1.2323,
+      "step": 4665
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.9396965031189545e-06,
+      "loss": 1.1626,
+      "step": 4670
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.9324304287530995e-06,
+      "loss": 1.1781,
+      "step": 4675
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.925162313727155e-06,
+      "loss": 1.2148,
+      "step": 4680
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.917892173947673e-06,
+      "loss": 1.1856,
+      "step": 4685
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.9106200253256375e-06,
+      "loss": 1.1836,
+      "step": 4690
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.903345883776424e-06,
+      "loss": 1.1742,
+      "step": 4695
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.896069765219774e-06,
+      "loss": 1.1929,
+      "step": 4700
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.8887916855797564e-06,
+      "loss": 1.2479,
+      "step": 4705
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.881511660784729e-06,
+      "loss": 1.2069,
+      "step": 4710
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.874229706767307e-06,
+      "loss": 1.2265,
+      "step": 4715
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.866945839464329e-06,
+      "loss": 1.2433,
+      "step": 4720
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.859660074816821e-06,
+      "loss": 1.2233,
+      "step": 4725
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.852372428769961e-06,
+      "loss": 1.2252,
+      "step": 4730
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.845082917273046e-06,
+      "loss": 1.2181,
+      "step": 4735
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.837791556279452e-06,
+      "loss": 1.2103,
+      "step": 4740
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.830498361746603e-06,
+      "loss": 1.1682,
+      "step": 4745
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.823203349635942e-06,
+      "loss": 1.1862,
+      "step": 4750
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.815906535912883e-06,
+      "loss": 1.1779,
+      "step": 4755
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.808607936546787e-06,
+      "loss": 1.2325,
+      "step": 4760
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.8013075675109174e-06,
+      "loss": 1.2146,
+      "step": 4765
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.794005444782418e-06,
+      "loss": 1.1979,
+      "step": 4770
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.786701584342265e-06,
+      "loss": 1.226,
+      "step": 4775
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.7793960021752424e-06,
+      "loss": 1.1729,
+      "step": 4780
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.7720887142698966e-06,
+      "loss": 1.2265,
+      "step": 4785
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.764779736618512e-06,
+      "loss": 1.1881,
+      "step": 4790
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.757469085217068e-06,
+      "loss": 1.2046,
+      "step": 4795
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.7501567760652075e-06,
+      "loss": 1.2291,
+      "step": 4800
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.742842825166206e-06,
+      "loss": 1.2548,
+      "step": 4805
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.735527248526922e-06,
+      "loss": 1.2161,
+      "step": 4810
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.728210062157784e-06,
+      "loss": 1.2241,
+      "step": 4815
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.720891282072734e-06,
+      "loss": 1.1633,
+      "step": 4820
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.713570924289209e-06,
+      "loss": 1.1876,
+      "step": 4825
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.706249004828091e-06,
+      "loss": 1.1748,
+      "step": 4830
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.6989255397136864e-06,
+      "loss": 1.15,
+      "step": 4835
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.691600544973682e-06,
+      "loss": 1.2395,
+      "step": 4840
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.684274036639114e-06,
+      "loss": 1.1762,
+      "step": 4845
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.676946030744326e-06,
+      "loss": 1.2414,
+      "step": 4850
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.669616543326946e-06,
+      "loss": 1.2158,
+      "step": 4855
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.6622855904278385e-06,
+      "loss": 1.1991,
+      "step": 4860
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.654953188091078e-06,
+      "loss": 1.2135,
+      "step": 4865
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.6476193523639114e-06,
+      "loss": 1.2314,
+      "step": 4870
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.640284099296723e-06,
+      "loss": 1.1973,
+      "step": 4875
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.632947444942998e-06,
+      "loss": 1.2072,
+      "step": 4880
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.625609405359287e-06,
+      "loss": 1.2057,
+      "step": 4885
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.618269996605174e-06,
+      "loss": 1.229,
+      "step": 4890
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.61092923474324e-06,
+      "loss": 1.2082,
+      "step": 4895
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.6035871358390256e-06,
+      "loss": 1.2337,
+      "step": 4900
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.596243715961e-06,
+      "loss": 1.2147,
+      "step": 4905
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.58889899118052e-06,
+      "loss": 1.2048,
+      "step": 4910
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.581552977571801e-06,
+      "loss": 1.16,
+      "step": 4915
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.574205691211878e-06,
+      "loss": 1.2421,
+      "step": 4920
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.566857148180572e-06,
+      "loss": 1.2817,
+      "step": 4925
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.559507364560454e-06,
+      "loss": 1.213,
+      "step": 4930
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.552156356436808e-06,
+      "loss": 1.2511,
+      "step": 4935
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.544804139897601e-06,
+      "loss": 1.183,
+      "step": 4940
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.537450731033447e-06,
+      "loss": 1.2229,
+      "step": 4945
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.530096145937561e-06,
+      "loss": 1.2265,
+      "step": 4950
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.522740400705739e-06,
+      "loss": 1.1829,
+      "step": 4955
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.515383511436314e-06,
+      "loss": 1.1802,
+      "step": 4960
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.508025494230124e-06,
+      "loss": 1.199,
+      "step": 4965
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.5006663651904735e-06,
+      "loss": 1.2155,
+      "step": 4970
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.4933061404231025e-06,
+      "loss": 1.1956,
+      "step": 4975
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.485944836036145e-06,
+      "loss": 1.1507,
+      "step": 4980
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.478582468140103e-06,
+      "loss": 1.1442,
+      "step": 4985
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.471219052847804e-06,
+      "loss": 1.2285,
+      "step": 4990
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.463854606274364e-06,
+      "loss": 1.2077,
+      "step": 4995
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.4564891445371635e-06,
+      "loss": 1.1715,
+      "step": 5000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.4491226837558e-06,
+      "loss": 1.1847,
+      "step": 5005
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.441755240052056e-06,
+      "loss": 1.2311,
+      "step": 5010
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.434386829549868e-06,
+      "loss": 1.2502,
+      "step": 5015
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.42701746837529e-06,
+      "loss": 1.1395,
+      "step": 5020
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.419647172656451e-06,
+      "loss": 1.2054,
+      "step": 5025
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.412275958523528e-06,
+      "loss": 1.2361,
+      "step": 5030
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.4049038421087115e-06,
+      "loss": 1.1857,
+      "step": 5035
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.397530839546162e-06,
+      "loss": 1.2728,
+      "step": 5040
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.3901569669719835e-06,
+      "loss": 1.1826,
+      "step": 5045
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.382782240524177e-06,
+      "loss": 1.1855,
+      "step": 5050
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.375406676342618e-06,
+      "loss": 1.2245,
+      "step": 5055
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.368030290569018e-06,
+      "loss": 1.1587,
+      "step": 5060
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.360653099346881e-06,
+      "loss": 1.2477,
+      "step": 5065
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.353275118821473e-06,
+      "loss": 1.2036,
+      "step": 5070
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.345896365139792e-06,
+      "loss": 1.1283,
+      "step": 5075
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.338516854450526e-06,
+      "loss": 1.1743,
+      "step": 5080
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.331136602904019e-06,
+      "loss": 1.1775,
+      "step": 5085
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.3237556266522394e-06,
+      "loss": 1.2209,
+      "step": 5090
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.316373941848738e-06,
+      "loss": 1.1405,
+      "step": 5095
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.308991564648617e-06,
+      "loss": 1.2079,
+      "step": 5100
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.301608511208495e-06,
+      "loss": 1.1826,
+      "step": 5105
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.294224797686471e-06,
+      "loss": 1.2145,
+      "step": 5110
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.286840440242087e-06,
+      "loss": 1.1783,
+      "step": 5115
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.279455455036296e-06,
+      "loss": 1.1832,
+      "step": 5120
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.272069858231424e-06,
+      "loss": 1.1996,
+      "step": 5125
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.2646836659911345e-06,
+      "loss": 1.2062,
+      "step": 5130
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.257296894480397e-06,
+      "loss": 1.1821,
+      "step": 5135
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.249909559865446e-06,
+      "loss": 1.2551,
+      "step": 5140
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.242521678313748e-06,
+      "loss": 1.2246,
+      "step": 5145
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.235133265993968e-06,
+      "loss": 1.1615,
+      "step": 5150
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.227744339075934e-06,
+      "loss": 1.1644,
+      "step": 5155
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.2203549137305975e-06,
+      "loss": 1.1886,
+      "step": 5160
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.212965006130003e-06,
+      "loss": 1.1512,
+      "step": 5165
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.205574632447248e-06,
+      "loss": 1.159,
+      "step": 5170
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.198183808856452e-06,
+      "loss": 1.2089,
+      "step": 5175
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.190792551532716e-06,
+      "loss": 1.2619,
+      "step": 5180
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.183400876652098e-06,
+      "loss": 1.1157,
+      "step": 5185
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.176008800391562e-06,
+      "loss": 1.2277,
+      "step": 5190
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.168616338928951e-06,
+      "loss": 1.1965,
+      "step": 5195
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.161223508442957e-06,
+      "loss": 1.1968,
+      "step": 5200
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.153830325113072e-06,
+      "loss": 1.2189,
+      "step": 5205
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.1464368051195654e-06,
+      "loss": 1.1644,
+      "step": 5210
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.139042964643443e-06,
+      "loss": 1.2008,
+      "step": 5215
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.131648819866411e-06,
+      "loss": 1.1781,
+      "step": 5220
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.124254386970838e-06,
+      "loss": 1.1887,
+      "step": 5225
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.116859682139731e-06,
+      "loss": 1.2135,
+      "step": 5230
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.109464721556685e-06,
+      "loss": 1.2185,
+      "step": 5235
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.102069521405859e-06,
+      "loss": 1.179,
+      "step": 5240
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.094674097871933e-06,
+      "loss": 1.1761,
+      "step": 5245
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.08727846714008e-06,
+      "loss": 1.162,
+      "step": 5250
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.0798826453959226e-06,
+      "loss": 1.2233,
+      "step": 5255
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.072486648825504e-06,
+      "loss": 1.1984,
+      "step": 5260
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.065090493615248e-06,
+      "loss": 1.2081,
+      "step": 5265
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.057694195951926e-06,
+      "loss": 1.2395,
+      "step": 5270
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.050297772022624e-06,
+      "loss": 1.2568,
+      "step": 5275
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.0429012380147e-06,
+      "loss": 1.2215,
+      "step": 5280
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.035504610115758e-06,
+      "loss": 1.244,
+      "step": 5285
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.0281079045136e-06,
+      "loss": 1.1802,
+      "step": 5290
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.020711137396206e-06,
+      "loss": 1.2338,
+      "step": 5295
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.013314324951686e-06,
+      "loss": 1.2284,
+      "step": 5300
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.00591748336825e-06,
+      "loss": 1.1705,
+      "step": 5305
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.998520628834173e-06,
+      "loss": 1.2354,
+      "step": 5310
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.9911237775377555e-06,
+      "loss": 1.0522,
+      "step": 5315
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.983726945667296e-06,
+      "loss": 0.9854,
+      "step": 5320
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.976330149411046e-06,
+      "loss": 1.0361,
+      "step": 5325
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.968933404957179e-06,
+      "loss": 1.0717,
+      "step": 5330
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.96153672849376e-06,
+      "loss": 1.0425,
+      "step": 5335
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.954140136208698e-06,
+      "loss": 0.9848,
+      "step": 5340
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.946743644289724e-06,
+      "loss": 1.0565,
+      "step": 5345
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.939347268924345e-06,
+      "loss": 1.0274,
+      "step": 5350
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.931951026299816e-06,
+      "loss": 1.0006,
+      "step": 5355
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.9245549326031025e-06,
+      "loss": 1.0687,
+      "step": 5360
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.917159004020837e-06,
+      "loss": 1.0681,
+      "step": 5365
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.909763256739299e-06,
+      "loss": 1.033,
+      "step": 5370
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.902367706944367e-06,
+      "loss": 1.0706,
+      "step": 5375
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8949723708214865e-06,
+      "loss": 1.0469,
+      "step": 5380
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.887577264555641e-06,
+      "loss": 1.0651,
+      "step": 5385
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.880182404331302e-06,
+      "loss": 1.0429,
+      "step": 5390
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.872787806332409e-06,
+      "loss": 1.0406,
+      "step": 5395
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.865393486742326e-06,
+      "loss": 0.9984,
+      "step": 5400
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.857999461743808e-06,
+      "loss": 0.9838,
+      "step": 5405
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.8506057475189665e-06,
+      "loss": 1.0564,
+      "step": 5410
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.843212360249229e-06,
+      "loss": 1.059,
+      "step": 5415
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.83581931611531e-06,
+      "loss": 0.9688,
+      "step": 5420
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.828426631297175e-06,
+      "loss": 1.0389,
+      "step": 5425
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.821034321973998e-06,
+      "loss": 1.0601,
+      "step": 5430
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.813642404324135e-06,
+      "loss": 1.0204,
+      "step": 5435
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.806250894525085e-06,
+      "loss": 1.0368,
+      "step": 5440
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.798859808753452e-06,
+      "loss": 1.0375,
+      "step": 5445
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.791469163184913e-06,
+      "loss": 1.0709,
+      "step": 5450
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.784078973994185e-06,
+      "loss": 1.0217,
+      "step": 5455
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.7766892573549795e-06,
+      "loss": 1.0082,
+      "step": 5460
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.76930002943998e-06,
+      "loss": 1.034,
+      "step": 5465
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.761911306420795e-06,
+      "loss": 0.9979,
+      "step": 5470
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.754523104467933e-06,
+      "loss": 1.0395,
+      "step": 5475
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.747135439750761e-06,
+      "loss": 1.0463,
+      "step": 5480
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.739748328437467e-06,
+      "loss": 1.0347,
+      "step": 5485
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.73236178669503e-06,
+      "loss": 1.031,
+      "step": 5490
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.7249758306891836e-06,
+      "loss": 1.0639,
+      "step": 5495
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.717590476584378e-06,
+      "loss": 1.0773,
+      "step": 5500
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.710205740543749e-06,
+      "loss": 1.0902,
+      "step": 5505
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.702821638729073e-06,
+      "loss": 1.0695,
+      "step": 5510
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.695438187300745e-06,
+      "loss": 1.0029,
+      "step": 5515
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.688055402417732e-06,
+      "loss": 1.0023,
+      "step": 5520
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.680673300237548e-06,
+      "loss": 1.053,
+      "step": 5525
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.6732918969162085e-06,
+      "loss": 0.9941,
+      "step": 5530
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.665911208608199e-06,
+      "loss": 1.0649,
+      "step": 5535
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.658531251466444e-06,
+      "loss": 1.057,
+      "step": 5540
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.651152041642263e-06,
+      "loss": 0.9844,
+      "step": 5545
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.643773595285343e-06,
+      "loss": 1.0367,
+      "step": 5550
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.636395928543699e-06,
+      "loss": 1.0047,
+      "step": 5555
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.629019057563641e-06,
+      "loss": 1.0491,
+      "step": 5560
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.621642998489737e-06,
+      "loss": 1.0534,
+      "step": 5565
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.614267767464778e-06,
+      "loss": 1.0255,
+      "step": 5570
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.606893380629742e-06,
+      "loss": 1.0086,
+      "step": 5575
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.599519854123758e-06,
+      "loss": 1.0521,
+      "step": 5580
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.592147204084078e-06,
+      "loss": 1.0184,
+      "step": 5585
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.58477544664603e-06,
+      "loss": 1.034,
+      "step": 5590
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.577404597942991e-06,
+      "loss": 1.0741,
+      "step": 5595
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.570034674106348e-06,
+      "loss": 0.9852,
+      "step": 5600
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.562665691265466e-06,
+      "loss": 1.0007,
+      "step": 5605
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.555297665547646e-06,
+      "loss": 1.003,
+      "step": 5610
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.547930613078102e-06,
+      "loss": 1.0468,
+      "step": 5615
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.540564549979911e-06,
+      "loss": 1.0503,
+      "step": 5620
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.533199492373987e-06,
+      "loss": 1.0404,
+      "step": 5625
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.525835456379045e-06,
+      "loss": 0.9809,
+      "step": 5630
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.518472458111562e-06,
+      "loss": 1.0032,
+      "step": 5635
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.511110513685744e-06,
+      "loss": 1.0327,
+      "step": 5640
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.503749639213491e-06,
+      "loss": 1.0155,
+      "step": 5645
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.496389850804365e-06,
+      "loss": 1.0591,
+      "step": 5650
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.489031164565545e-06,
+      "loss": 1.0272,
+      "step": 5655
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.481673596601802e-06,
+      "loss": 1.0068,
+      "step": 5660
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.474317163015458e-06,
+      "loss": 1.0255,
+      "step": 5665
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.4669618799063516e-06,
+      "loss": 1.0601,
+      "step": 5670
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.459607763371806e-06,
+      "loss": 1.0159,
+      "step": 5675
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.452254829506589e-06,
+      "loss": 1.0418,
+      "step": 5680
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.444903094402882e-06,
+      "loss": 1.002,
+      "step": 5685
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.4375525741502406e-06,
+      "loss": 1.0388,
+      "step": 5690
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.430203284835565e-06,
+      "loss": 1.0215,
+      "step": 5695
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.422855242543056e-06,
+      "loss": 1.034,
+      "step": 5700
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.415508463354191e-06,
+      "loss": 1.0031,
+      "step": 5705
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.408162963347682e-06,
+      "loss": 0.9963,
+      "step": 5710
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.400818758599436e-06,
+      "loss": 1.0124,
+      "step": 5715
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.3934758651825334e-06,
+      "loss": 0.9911,
+      "step": 5720
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.386134299167178e-06,
+      "loss": 1.0429,
+      "step": 5725
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.378794076620672e-06,
+      "loss": 1.0365,
+      "step": 5730
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.371455213607378e-06,
+      "loss": 1.0009,
+      "step": 5735
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.3641177261886795e-06,
+      "loss": 1.0523,
+      "step": 5740
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.356781630422955e-06,
+      "loss": 1.0195,
+      "step": 5745
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.349446942365529e-06,
+      "loss": 1.019,
+      "step": 5750
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.342113678068655e-06,
+      "loss": 0.989,
+      "step": 5755
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.334781853581463e-06,
+      "loss": 1.0186,
+      "step": 5760
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.327451484949936e-06,
+      "loss": 0.9944,
+      "step": 5765
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.320122588216868e-06,
+      "loss": 1.0536,
+      "step": 5770
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.312795179421836e-06,
+      "loss": 1.0433,
+      "step": 5775
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.305469274601156e-06,
+      "loss": 1.0048,
+      "step": 5780
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.298144889787855e-06,
+      "loss": 1.0851,
+      "step": 5785
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.290822041011632e-06,
+      "loss": 1.047,
+      "step": 5790
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.283500744298825e-06,
+      "loss": 1.0821,
+      "step": 5795
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.276181015672375e-06,
+      "loss": 1.0109,
+      "step": 5800
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.268862871151792e-06,
+      "loss": 1.0063,
+      "step": 5805
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.261546326753119e-06,
+      "loss": 0.98,
+      "step": 5810
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.254231398488897e-06,
+      "loss": 0.9981,
+      "step": 5815
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.246918102368128e-06,
+      "loss": 1.0593,
+      "step": 5820
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.2396064543962445e-06,
+      "loss": 0.9974,
+      "step": 5825
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.232296470575071e-06,
+      "loss": 1.0358,
+      "step": 5830
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.2249881669027904e-06,
+      "loss": 1.0716,
+      "step": 5835
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.217681559373909e-06,
+      "loss": 1.0109,
+      "step": 5840
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.2103766639792185e-06,
+      "loss": 1.0453,
+      "step": 5845
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.203073496705765e-06,
+      "loss": 0.9672,
+      "step": 5850
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.195772073536814e-06,
+      "loss": 1.0695,
+      "step": 5855
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.188472410451814e-06,
+      "loss": 0.9622,
+      "step": 5860
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.181174523426359e-06,
+      "loss": 0.9904,
+      "step": 5865
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.1738784284321554e-06,
+      "loss": 1.0098,
+      "step": 5870
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.166584141436992e-06,
+      "loss": 1.0066,
+      "step": 5875
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.159291678404697e-06,
+      "loss": 1.0213,
+      "step": 5880
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.1520010552951075e-06,
+      "loss": 1.0107,
+      "step": 5885
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.144712288064037e-06,
+      "loss": 1.0401,
+      "step": 5890
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.137425392663231e-06,
+      "loss": 1.0016,
+      "step": 5895
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.130140385040346e-06,
+      "loss": 1.0589,
+      "step": 5900
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.122857281138902e-06,
+      "loss": 1.061,
+      "step": 5905
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.115576096898253e-06,
+      "loss": 0.9996,
+      "step": 5910
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.108296848253552e-06,
+      "loss": 1.0077,
+      "step": 5915
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.101019551135718e-06,
+      "loss": 1.0388,
+      "step": 5920
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.093744221471396e-06,
+      "loss": 1.0023,
+      "step": 5925
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.086470875182929e-06,
+      "loss": 1.0066,
+      "step": 5930
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.079199528188315e-06,
+      "loss": 0.9858,
+      "step": 5935
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.0719301964011774e-06,
+      "loss": 0.9718,
+      "step": 5940
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.064662895730732e-06,
+      "loss": 1.0067,
+      "step": 5945
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.057397642081747e-06,
+      "loss": 0.9826,
+      "step": 5950
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.050134451354511e-06,
+      "loss": 1.0515,
+      "step": 5955
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.0428733394448005e-06,
+      "loss": 1.0115,
+      "step": 5960
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.035614322243836e-06,
+      "loss": 0.9815,
+      "step": 5965
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.02835741563826e-06,
+      "loss": 0.9842,
+      "step": 5970
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.021102635510093e-06,
+      "loss": 1.0193,
+      "step": 5975
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.013849997736703e-06,
+      "loss": 1.0357,
+      "step": 5980
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.006599518190768e-06,
+      "loss": 1.0696,
+      "step": 5985
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.999351212740242e-06,
+      "loss": 0.9836,
+      "step": 5990
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.992105097248324e-06,
+      "loss": 1.0348,
+      "step": 5995
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.984861187573416e-06,
+      "loss": 0.9809,
+      "step": 6000
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.977619499569096e-06,
+      "loss": 1.0009,
+      "step": 6005
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.970380049084081e-06,
+      "loss": 1.0564,
+      "step": 6010
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.9631428519621845e-06,
+      "loss": 0.9941,
+      "step": 6015
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.9559079240422916e-06,
+      "loss": 1.0416,
+      "step": 6020
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.948675281158329e-06,
+      "loss": 1.0269,
+      "step": 6025
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.94144493913921e-06,
+      "loss": 1.0337,
+      "step": 6030
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.9342169138088184e-06,
+      "loss": 1.0,
+      "step": 6035
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.92699122098597e-06,
+      "loss": 1.0053,
+      "step": 6040
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.91976787648437e-06,
+      "loss": 1.0512,
+      "step": 6045
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.91254689611259e-06,
+      "loss": 0.9962,
+      "step": 6050
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.905328295674026e-06,
+      "loss": 1.0107,
+      "step": 6055
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.898112090966861e-06,
+      "loss": 1.0099,
+      "step": 6060
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.89089829778404e-06,
+      "loss": 0.9886,
+      "step": 6065
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.883686931913229e-06,
+      "loss": 0.9906,
+      "step": 6070
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.876478009136781e-06,
+      "loss": 0.9999,
+      "step": 6075
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.869271545231704e-06,
+      "loss": 0.9761,
+      "step": 6080
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.862067555969621e-06,
+      "loss": 0.9534,
+      "step": 6085
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.854866057116742e-06,
+      "loss": 1.0182,
+      "step": 6090
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.847667064433825e-06,
+      "loss": 1.0035,
+      "step": 6095
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.840470593676148e-06,
+      "loss": 1.0189,
+      "step": 6100
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.8332766605934624e-06,
+      "loss": 0.9904,
+      "step": 6105
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.82608528092997e-06,
+      "loss": 1.0312,
+      "step": 6110
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.818896470424283e-06,
+      "loss": 1.0193,
+      "step": 6115
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.8117102448093922e-06,
+      "loss": 1.0401,
+      "step": 6120
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.8045266198126314e-06,
+      "loss": 1.0227,
+      "step": 6125
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.797345611155642e-06,
+      "loss": 1.0284,
+      "step": 6130
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.7901672345543366e-06,
+      "loss": 0.9987,
+      "step": 6135
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.7829915057188727e-06,
+      "loss": 0.9681,
+      "step": 6140
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.775818440353609e-06,
+      "loss": 1.0071,
+      "step": 6145
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.7686480541570796e-06,
+      "loss": 1.0211,
+      "step": 6150
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.7614803628219498e-06,
+      "loss": 0.9708,
+      "step": 6155
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.754315382034991e-06,
+      "loss": 0.9845,
+      "step": 6160
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.7471531274770394e-06,
+      "loss": 1.0052,
+      "step": 6165
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.7399936148229676e-06,
+      "loss": 1.0226,
+      "step": 6170
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.7328368597416477e-06,
+      "loss": 1.0353,
+      "step": 6175
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.7256828778959116e-06,
+      "loss": 0.9976,
+      "step": 6180
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.7185316849425272e-06,
+      "loss": 0.9702,
+      "step": 6185
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.711383296532158e-06,
+      "loss": 1.0527,
+      "step": 6190
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.7042377283093272e-06,
+      "loss": 1.0742,
+      "step": 6195
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.697094995912389e-06,
+      "loss": 1.0034,
+      "step": 6200
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.6899551149734893e-06,
+      "loss": 1.0184,
+      "step": 6205
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.6828181011185326e-06,
+      "loss": 1.0611,
+      "step": 6210
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.6756839699671515e-06,
+      "loss": 1.0041,
+      "step": 6215
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.668552737132668e-06,
+      "loss": 1.0123,
+      "step": 6220
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.6614244182220608e-06,
+      "loss": 0.9527,
+      "step": 6225
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.654299028835932e-06,
+      "loss": 1.0069,
+      "step": 6230
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.6471765845684715e-06,
+      "loss": 1.0283,
+      "step": 6235
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.640057101007425e-06,
+      "loss": 1.009,
+      "step": 6240
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.632940593734058e-06,
+      "loss": 1.0457,
+      "step": 6245
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.625827078323122e-06,
+      "loss": 0.9835,
+      "step": 6250
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.618716570342821e-06,
+      "loss": 1.0182,
+      "step": 6255
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.611609085354776e-06,
+      "loss": 0.9658,
+      "step": 6260
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.6045046389139922e-06,
+      "loss": 1.0461,
+      "step": 6265
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.5974032465688286e-06,
+      "loss": 1.0559,
+      "step": 6270
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.5903049238609555e-06,
+      "loss": 1.0029,
+      "step": 6275
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.583209686325327e-06,
+      "loss": 0.9966,
+      "step": 6280
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.5761175494901466e-06,
+      "loss": 1.0301,
+      "step": 6285
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.5690285288768284e-06,
+      "loss": 0.9866,
+      "step": 6290
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.56194263999997e-06,
+      "loss": 1.0016,
+      "step": 6295
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.554859898367312e-06,
+      "loss": 1.035,
+      "step": 6300
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.547780319479709e-06,
+      "loss": 0.9535,
+      "step": 6305
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.5407039188310933e-06,
+      "loss": 1.0038,
+      "step": 6310
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.5336307119084425e-06,
+      "loss": 0.9908,
+      "step": 6315
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.526560714191745e-06,
+      "loss": 0.9635,
+      "step": 6320
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.5194939411539604e-06,
+      "loss": 1.0042,
+      "step": 6325
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.5124304082609978e-06,
+      "loss": 1.0142,
+      "step": 6330
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.5053701309716724e-06,
+      "loss": 1.0364,
+      "step": 6335
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.498313124737674e-06,
+      "loss": 0.9686,
+      "step": 6340
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.4912594050035346e-06,
+      "loss": 1.0308,
+      "step": 6345
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.4842089872065915e-06,
+      "loss": 0.9767,
+      "step": 6350
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.4771618867769574e-06,
+      "loss": 1.0148,
+      "step": 6355
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.4701181191374855e-06,
+      "loss": 1.0103,
+      "step": 6360
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.4630776997037324e-06,
+      "loss": 1.0346,
+      "step": 6365
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.45604064388393e-06,
+      "loss": 1.0342,
+      "step": 6370
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.4490069670789468e-06,
+      "loss": 0.9953,
+      "step": 6375
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.4419766846822553e-06,
+      "loss": 0.9898,
+      "step": 6380
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.434949812079903e-06,
+      "loss": 1.0205,
+      "step": 6385
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.4279263646504702e-06,
+      "loss": 0.9656,
+      "step": 6390
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.420906357765046e-06,
+      "loss": 1.0617,
+      "step": 6395
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.4138898067871864e-06,
+      "loss": 1.0423,
+      "step": 6400
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.406876727072884e-06,
+      "loss": 0.9904,
+      "step": 6405
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.3998671339705362e-06,
+      "loss": 0.9708,
+      "step": 6410
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.3928610428209097e-06,
+      "loss": 1.0459,
+      "step": 6415
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.3858584689571038e-06,
+      "loss": 1.0256,
+      "step": 6420
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.3788594277045233e-06,
+      "loss": 0.9976,
+      "step": 6425
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.3718639343808414e-06,
+      "loss": 1.0649,
+      "step": 6430
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.364872004295966e-06,
+      "loss": 1.0167,
+      "step": 6435
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.3578836527520078e-06,
+      "loss": 1.0348,
+      "step": 6440
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.3508988950432418e-06,
+      "loss": 0.9988,
+      "step": 6445
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.3439177464560833e-06,
+      "loss": 1.0256,
+      "step": 6450
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.336940222269045e-06,
+      "loss": 0.9936,
+      "step": 6455
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.3299663377527087e-06,
+      "loss": 0.9776,
+      "step": 6460
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.3229961081696917e-06,
+      "loss": 0.9983,
+      "step": 6465
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.3160295487746096e-06,
+      "loss": 0.9799,
+      "step": 6470
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.309066674814049e-06,
+      "loss": 1.0058,
+      "step": 6475
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.3021075015265277e-06,
+      "loss": 1.0097,
+      "step": 6480
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.295152044142466e-06,
+      "loss": 1.035,
+      "step": 6485
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.2882003178841536e-06,
+      "loss": 1.0444,
+      "step": 6490
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.2812523379657103e-06,
+      "loss": 1.0425,
+      "step": 6495
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.274308119593059e-06,
+      "loss": 1.023,
+      "step": 6500
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.267367677963892e-06,
+      "loss": 1.0267,
+      "step": 6505
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.2604310282676332e-06,
+      "loss": 1.0291,
+      "step": 6510
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.253498185685409e-06,
+      "loss": 0.9999,
+      "step": 6515
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.2465691653900144e-06,
+      "loss": 1.0131,
+      "step": 6520
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.2396439825458792e-06,
+      "loss": 1.0255,
+      "step": 6525
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.2327226523090328e-06,
+      "loss": 0.9865,
+      "step": 6530
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.2258051898270766e-06,
+      "loss": 0.9833,
+      "step": 6535
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.218891610239142e-06,
+      "loss": 1.0122,
+      "step": 6540
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.211981928675867e-06,
+      "loss": 1.014,
+      "step": 6545
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.2050761602593568e-06,
+      "loss": 0.9609,
+      "step": 6550
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.1981743201031535e-06,
+      "loss": 1.0376,
+      "step": 6555
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.191276423312202e-06,
+      "loss": 1.0241,
+      "step": 6560
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.1843824849828144e-06,
+      "loss": 1.0482,
+      "step": 6565
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.1774925202026425e-06,
+      "loss": 0.983,
+      "step": 6570
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.1706065440506418e-06,
+      "loss": 0.9573,
+      "step": 6575
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.1637245715970364e-06,
+      "loss": 1.0136,
+      "step": 6580
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.156846617903292e-06,
+      "loss": 1.0236,
+      "step": 6585
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.1499726980220734e-06,
+      "loss": 1.0485,
+      "step": 6590
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.1431028269972215e-06,
+      "loss": 1.0364,
+      "step": 6595
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.1362370198637147e-06,
+      "loss": 0.9959,
+      "step": 6600
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.1293752916476387e-06,
+      "loss": 1.022,
+      "step": 6605
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.1225176573661513e-06,
+      "loss": 1.0229,
+      "step": 6610
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.1156641320274494e-06,
+      "loss": 1.0539,
+      "step": 6615
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.1088147306307385e-06,
+      "loss": 1.029,
+      "step": 6620
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.1019694681662e-06,
+      "loss": 1.0065,
+      "step": 6625
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.0951283596149547e-06,
+      "loss": 1.0124,
+      "step": 6630
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.088291419949033e-06,
+      "loss": 1.0286,
+      "step": 6635
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.081458664131344e-06,
+      "loss": 0.978,
+      "step": 6640
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.0746301071156363e-06,
+      "loss": 1.0259,
+      "step": 6645
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.067805763846473e-06,
+      "loss": 1.0355,
+      "step": 6650
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.0609856492591937e-06,
+      "loss": 0.9806,
+      "step": 6655
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.0541697782798806e-06,
+      "loss": 0.9485,
+      "step": 6660
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.0473581658253337e-06,
+      "loss": 1.053,
+      "step": 6665
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.0405508268030295e-06,
+      "loss": 1.0301,
+      "step": 6670
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.033747776111092e-06,
+      "loss": 0.9644,
+      "step": 6675
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.026949028638264e-06,
+      "loss": 0.9973,
+      "step": 6680
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.020154599263865e-06,
+      "loss": 0.931,
+      "step": 6685
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.013364502857766e-06,
+      "loss": 1.0684,
+      "step": 6690
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.006578754280358e-06,
+      "loss": 1.04,
+      "step": 6695
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.999797368382513e-06,
+      "loss": 1.0232,
+      "step": 6700
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9930203600055596e-06,
+      "loss": 0.9895,
+      "step": 6705
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.986247743981238e-06,
+      "loss": 0.9993,
+      "step": 6710
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.979479535131684e-06,
+      "loss": 0.969,
+      "step": 6715
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.9727157482693835e-06,
+      "loss": 1.0122,
+      "step": 6720
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.965956398197146e-06,
+      "loss": 1.0306,
+      "step": 6725
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.959201499708072e-06,
+      "loss": 1.0005,
+      "step": 6730
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.9524510675855155e-06,
+      "loss": 1.0664,
+      "step": 6735
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.9457051166030616e-06,
+      "loss": 1.0063,
+      "step": 6740
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.938963661524483e-06,
+      "loss": 1.022,
+      "step": 6745
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.932226717103716e-06,
+      "loss": 0.9959,
+      "step": 6750
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.9254942980848245e-06,
+      "loss": 1.0167,
+      "step": 6755
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.9187664192019676e-06,
+      "loss": 1.0054,
+      "step": 6760
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.9120430951793677e-06,
+      "loss": 0.9725,
+      "step": 6765
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.9053243407312835e-06,
+      "loss": 1.0028,
+      "step": 6770
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.898610170561964e-06,
+      "loss": 0.9773,
+      "step": 6775
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8919005993656357e-06,
+      "loss": 1.0138,
+      "step": 6780
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8851956418264523e-06,
+      "loss": 1.0131,
+      "step": 6785
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8784953126184704e-06,
+      "loss": 0.9964,
+      "step": 6790
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.871799626405625e-06,
+      "loss": 1.0025,
+      "step": 6795
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8651085978416798e-06,
+      "loss": 1.0074,
+      "step": 6800
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.858422241570214e-06,
+      "loss": 0.968,
+      "step": 6805
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.851740572224575e-06,
+      "loss": 1.0276,
+      "step": 6810
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8450636044278535e-06,
+      "loss": 1.0326,
+      "step": 6815
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.838391352792855e-06,
+      "loss": 0.9835,
+      "step": 6820
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8317238319220564e-06,
+      "loss": 0.9922,
+      "step": 6825
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.825061056407591e-06,
+      "loss": 0.9719,
+      "step": 6830
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8184030408311947e-06,
+      "loss": 0.9973,
+      "step": 6835
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.811749799764198e-06,
+      "loss": 1.0141,
+      "step": 6840
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8051013477674736e-06,
+      "loss": 1.0195,
+      "step": 6845
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.7984576993914143e-06,
+      "loss": 1.0166,
+      "step": 6850
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.791818869175906e-06,
+      "loss": 1.0177,
+      "step": 6855
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.7851848716502815e-06,
+      "loss": 1.0475,
+      "step": 6860
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.7785557213333047e-06,
+      "loss": 0.9932,
+      "step": 6865
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.771931432733126e-06,
+      "loss": 1.0372,
+      "step": 6870
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.765312020347256e-06,
+      "loss": 1.0422,
+      "step": 6875
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.758697498662538e-06,
+      "loss": 0.9787,
+      "step": 6880
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.7520878821551063e-06,
+      "loss": 0.9881,
+      "step": 6885
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.745483185290363e-06,
+      "loss": 0.9499,
+      "step": 6890
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.7388834225229445e-06,
+      "loss": 1.0188,
+      "step": 6895
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.732288608296685e-06,
+      "loss": 1.0068,
+      "step": 6900
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.7256987570445925e-06,
+      "loss": 1.03,
+      "step": 6905
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.7191138831888113e-06,
+      "loss": 1.0066,
+      "step": 6910
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.71253400114059e-06,
+      "loss": 0.9573,
+      "step": 6915
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.7059591253002595e-06,
+      "loss": 1.0329,
+      "step": 6920
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.699389270057186e-06,
+      "loss": 1.0484,
+      "step": 6925
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.692824449789756e-06,
+      "loss": 0.9456,
+      "step": 6930
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.68626467886533e-06,
+      "loss": 1.0412,
+      "step": 6935
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.67970997164022e-06,
+      "loss": 1.0089,
+      "step": 6940
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.67316034245966e-06,
+      "loss": 1.0235,
+      "step": 6945
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.666615805657762e-06,
+      "loss": 1.0095,
+      "step": 6950
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.6600763755575037e-06,
+      "loss": 1.0048,
+      "step": 6955
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.653542066470678e-06,
+      "loss": 0.9806,
+      "step": 6960
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.6470128926978717e-06,
+      "loss": 1.0156,
+      "step": 6965
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.640488868528438e-06,
+      "loss": 1.0357,
+      "step": 6970
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.633970008240453e-06,
+      "loss": 1.0101,
+      "step": 6975
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.627456326100698e-06,
+      "loss": 1.0412,
+      "step": 6980
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.620947836364617e-06,
+      "loss": 0.9919,
+      "step": 6985
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.614444553276289e-06,
+      "loss": 0.9888,
+      "step": 6990
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.607946491068406e-06,
+      "loss": 1.0312,
+      "step": 6995
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.6014536639622245e-06,
+      "loss": 1.0067,
+      "step": 7000
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.594966086167551e-06,
+      "loss": 1.0137,
+      "step": 7005
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.5884837718826972e-06,
+      "loss": 0.999,
+      "step": 7010
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.5820067352944612e-06,
+      "loss": 1.0558,
+      "step": 7015
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.5755349905780892e-06,
+      "loss": 0.9686,
+      "step": 7020
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.5690685518972447e-06,
+      "loss": 1.0499,
+      "step": 7025
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.562607433403976e-06,
+      "loss": 1.0096,
+      "step": 7030
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.5561516492386943e-06,
+      "loss": 1.0189,
+      "step": 7035
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.5497012135301312e-06,
+      "loss": 1.0226,
+      "step": 7040
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.543256140395317e-06,
+      "loss": 0.9886,
+      "step": 7045
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.5368164439395434e-06,
+      "loss": 1.0389,
+      "step": 7050
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.5303821382563322e-06,
+      "loss": 0.9977,
+      "step": 7055
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.5239532374274145e-06,
+      "loss": 1.0542,
+      "step": 7060
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.517529755522685e-06,
+      "loss": 0.9817,
+      "step": 7065
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.5111117066001867e-06,
+      "loss": 1.0289,
+      "step": 7070
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.5046991047060643e-06,
+      "loss": 0.961,
+      "step": 7075
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.498291963874548e-06,
+      "loss": 1.0004,
+      "step": 7080
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.4918902981279125e-06,
+      "loss": 1.0639,
+      "step": 7085
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.4854941214764496e-06,
+      "loss": 1.0327,
+      "step": 7090
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.479103447918442e-06,
+      "loss": 1.0185,
+      "step": 7095
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.4727182914401243e-06,
+      "loss": 1.0146,
+      "step": 7100
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.4663386660156607e-06,
+      "loss": 1.0032,
+      "step": 7105
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.459964585607107e-06,
+      "loss": 1.0132,
+      "step": 7110
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.453596064164383e-06,
+      "loss": 0.9677,
+      "step": 7115
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.447233115625249e-06,
+      "loss": 1.0285,
+      "step": 7120
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.4408757539152587e-06,
+      "loss": 1.0293,
+      "step": 7125
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.4345239929477498e-06,
+      "loss": 1.0272,
+      "step": 7130
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.4281778466237925e-06,
+      "loss": 0.9806,
+      "step": 7135
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.4218373288321757e-06,
+      "loss": 0.9967,
+      "step": 7140
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.41550245344937e-06,
+      "loss": 0.993,
+      "step": 7145
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.409173234339493e-06,
+      "loss": 0.9943,
+      "step": 7150
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.402849685354285e-06,
+      "loss": 0.9987,
+      "step": 7155
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.39653182033308e-06,
+      "loss": 0.9289,
+      "step": 7160
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.3902196531027676e-06,
+      "loss": 1.0155,
+      "step": 7165
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.3839131974777737e-06,
+      "loss": 0.9853,
+      "step": 7170
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.3776124672600183e-06,
+      "loss": 0.9608,
+      "step": 7175
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.371317476238891e-06,
+      "loss": 1.0193,
+      "step": 7180
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.3650282381912287e-06,
+      "loss": 1.004,
+      "step": 7185
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.3587447668812674e-06,
+      "loss": 1.0586,
+      "step": 7190
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.352467076060631e-06,
+      "loss": 0.9975,
+      "step": 7195
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.3461951794682873e-06,
+      "loss": 1.0465,
+      "step": 7200
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.3399290908305235e-06,
+      "loss": 0.977,
+      "step": 7205
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.3336688238609195e-06,
+      "loss": 1.0139,
+      "step": 7210
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.3274143922603094e-06,
+      "loss": 1.0116,
+      "step": 7215
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.321165809716762e-06,
+      "loss": 0.9869,
+      "step": 7220
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.3149230899055408e-06,
+      "loss": 1.015,
+      "step": 7225
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.3086862464890776e-06,
+      "loss": 1.0494,
+      "step": 7230
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.3024552931169485e-06,
+      "loss": 1.0413,
+      "step": 7235
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.296230243425834e-06,
+      "loss": 0.9762,
+      "step": 7240
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.290011111039499e-06,
+      "loss": 1.0236,
+      "step": 7245
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.2837979095687534e-06,
+      "loss": 0.9891,
+      "step": 7250
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.277590652611428e-06,
+      "loss": 1.0041,
+      "step": 7255
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.2713893537523476e-06,
+      "loss": 1.0105,
+      "step": 7260
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.2651940265632943e-06,
+      "loss": 1.0194,
+      "step": 7265
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.2590046846029794e-06,
+      "loss": 1.062,
+      "step": 7270
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.2528213414170215e-06,
+      "loss": 0.9766,
+      "step": 7275
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.246644010537901e-06,
+      "loss": 1.0246,
+      "step": 7280
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.2404727054849513e-06,
+      "loss": 0.9829,
+      "step": 7285
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.2343074397643095e-06,
+      "loss": 1.0424,
+      "step": 7290
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.2281482268688963e-06,
+      "loss": 1.0679,
+      "step": 7295
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.2219950802783922e-06,
+      "loss": 1.0045,
+      "step": 7300
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.2158480134591927e-06,
+      "loss": 1.0126,
+      "step": 7305
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.2097070398643937e-06,
+      "loss": 0.9906,
+      "step": 7310
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.2035721729337524e-06,
+      "loss": 0.9894,
+      "step": 7315
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.1974434260936637e-06,
+      "loss": 1.0052,
+      "step": 7320
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.191320812757128e-06,
+      "loss": 0.9687,
+      "step": 7325
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.1852043463237184e-06,
+      "loss": 0.9843,
+      "step": 7330
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.1790940401795617e-06,
+      "loss": 1.0324,
+      "step": 7335
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.1729899076972967e-06,
+      "loss": 1.0006,
+      "step": 7340
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.166891962236057e-06,
+      "loss": 0.9852,
+      "step": 7345
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.16080021714143e-06,
+      "loss": 1.0015,
+      "step": 7350
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.1547146857454344e-06,
+      "loss": 1.0451,
+      "step": 7355
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.1486353813664946e-06,
+      "loss": 0.9813,
+      "step": 7360
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.1425623173094e-06,
+      "loss": 0.9696,
+      "step": 7365
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.1364955068652905e-06,
+      "loss": 1.0468,
+      "step": 7370
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.1304349633116135e-06,
+      "loss": 0.9883,
+      "step": 7375
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.1243806999121007e-06,
+      "loss": 1.0037,
+      "step": 7380
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.1183327299167443e-06,
+      "loss": 1.0428,
+      "step": 7385
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.112291066561763e-06,
+      "loss": 1.0181,
+      "step": 7390
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.1062557230695665e-06,
+      "loss": 0.9839,
+      "step": 7395
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.1002267126487414e-06,
+      "loss": 0.9951,
+      "step": 7400
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.0942040484940075e-06,
+      "loss": 1.0095,
+      "step": 7405
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.0881877437862006e-06,
+      "loss": 1.0413,
+      "step": 7410
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.0821778116922362e-06,
+      "loss": 1.0079,
+      "step": 7415
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.076174265365081e-06,
+      "loss": 1.0003,
+      "step": 7420
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.0701771179437323e-06,
+      "loss": 1.0734,
+      "step": 7425
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.0641863825531756e-06,
+      "loss": 0.9623,
+      "step": 7430
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.0582020723043706e-06,
+      "loss": 1.0149,
+      "step": 7435
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.05222420029421e-06,
+      "loss": 0.9964,
+      "step": 7440
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.046252779605497e-06,
+      "loss": 0.9638,
+      "step": 7445
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.04028782330692e-06,
+      "loss": 0.9799,
+      "step": 7450
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.034329344453014e-06,
+      "loss": 1.0072,
+      "step": 7455
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.0283773560841425e-06,
+      "loss": 1.0516,
+      "step": 7460
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.0224318712264623e-06,
+      "loss": 1.039,
+      "step": 7465
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.0164929028918943e-06,
+      "loss": 1.0007,
+      "step": 7470
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.010560464078105e-06,
+      "loss": 1.0043,
+      "step": 7475
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.0046345677684626e-06,
+      "loss": 0.9936,
+      "step": 7480
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.998715226932024e-06,
+      "loss": 1.0092,
+      "step": 7485
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9928024545234935e-06,
+      "loss": 1.009,
+      "step": 7490
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.986896263483202e-06,
+      "loss": 1.0476,
+      "step": 7495
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9809966667370793e-06,
+      "loss": 1.0099,
+      "step": 7500
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.975103677196619e-06,
+      "loss": 1.0572,
+      "step": 7505
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.969217307758857e-06,
+      "loss": 1.0021,
+      "step": 7510
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9633375713063442e-06,
+      "loss": 0.9912,
+      "step": 7515
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9574644807071063e-06,
+      "loss": 0.9782,
+      "step": 7520
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.951598048814634e-06,
+      "loss": 1.0154,
+      "step": 7525
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.945738288467839e-06,
+      "loss": 1.0426,
+      "step": 7530
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.939885212491031e-06,
+      "loss": 1.0245,
+      "step": 7535
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9340388336938976e-06,
+      "loss": 0.9687,
+      "step": 7540
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9281991648714613e-06,
+      "loss": 0.9701,
+      "step": 7545
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.922366218804067e-06,
+      "loss": 0.977,
+      "step": 7550
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9165400082573393e-06,
+      "loss": 0.9886,
+      "step": 7555
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.910720545982169e-06,
+      "loss": 1.0078,
+      "step": 7560
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9049078447146729e-06,
+      "loss": 0.9872,
+      "step": 7565
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.8991019171761714e-06,
+      "loss": 1.0382,
+      "step": 7570
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.8933027760731642e-06,
+      "loss": 0.9738,
+      "step": 7575
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.8875104340972939e-06,
+      "loss": 0.9879,
+      "step": 7580
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.8817249039253277e-06,
+      "loss": 0.9897,
+      "step": 7585
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.8759461982191206e-06,
+      "loss": 0.9712,
+      "step": 7590
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.870174329625592e-06,
+      "loss": 1.0142,
+      "step": 7595
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.8644093107767042e-06,
+      "loss": 1.0416,
+      "step": 7600
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.8586511542894192e-06,
+      "loss": 1.002,
+      "step": 7605
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.8528998727656899e-06,
+      "loss": 0.9846,
+      "step": 7610
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.8471554787924162e-06,
+      "loss": 1.063,
+      "step": 7615
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.8414179849414255e-06,
+      "loss": 0.9915,
+      "step": 7620
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.8356874037694483e-06,
+      "loss": 1.0099,
+      "step": 7625
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.8299637478180798e-06,
+      "loss": 1.0092,
+      "step": 7630
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.8242470296137644e-06,
+      "loss": 1.0749,
+      "step": 7635
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.8185372616677633e-06,
+      "loss": 0.9846,
+      "step": 7640
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.8128344564761207e-06,
+      "loss": 0.9644,
+      "step": 7645
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.8071386265196506e-06,
+      "loss": 0.9569,
+      "step": 7650
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.801449784263895e-06,
+      "loss": 1.0062,
+      "step": 7655
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.795767942159104e-06,
+      "loss": 1.0086,
+      "step": 7660
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.790093112640212e-06,
+      "loss": 0.9769,
+      "step": 7665
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7844253081268003e-06,
+      "loss": 0.9795,
+      "step": 7670
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7787645410230815e-06,
+      "loss": 1.0027,
+      "step": 7675
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.773110823717862e-06,
+      "loss": 1.0327,
+      "step": 7680
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.767464168584519e-06,
+      "loss": 1.0469,
+      "step": 7685
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7618245879809797e-06,
+      "loss": 1.0232,
+      "step": 7690
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7561920942496807e-06,
+      "loss": 1.0175,
+      "step": 7695
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7505666997175564e-06,
+      "loss": 0.9814,
+      "step": 7700
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7449484166959984e-06,
+      "loss": 1.0214,
+      "step": 7705
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7393372574808354e-06,
+      "loss": 0.9433,
+      "step": 7710
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7337332343523095e-06,
+      "loss": 1.0303,
+      "step": 7715
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.728136359575039e-06,
+      "loss": 0.9507,
+      "step": 7720
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7225466453980044e-06,
+      "loss": 0.9991,
+      "step": 7725
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7169641040545098e-06,
+      "loss": 0.9836,
+      "step": 7730
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7113887477621615e-06,
+      "loss": 0.996,
+      "step": 7735
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7058205887228458e-06,
+      "loss": 0.9772,
+      "step": 7740
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7002596391226905e-06,
+      "loss": 1.0214,
+      "step": 7745
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6947059111320535e-06,
+      "loss": 0.9767,
+      "step": 7750
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6891594169054797e-06,
+      "loss": 1.0001,
+      "step": 7755
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6836201685816878e-06,
+      "loss": 1.0066,
+      "step": 7760
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.678088178283539e-06,
+      "loss": 1.0153,
+      "step": 7765
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6725634581180066e-06,
+      "loss": 0.9883,
+      "step": 7770
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6670460201761523e-06,
+      "loss": 0.9997,
+      "step": 7775
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6615358765331075e-06,
+      "loss": 0.9908,
+      "step": 7780
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.65603303924803e-06,
+      "loss": 1.0155,
+      "step": 7785
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6505375203640961e-06,
+      "loss": 1.0399,
+      "step": 7790
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6450493319084581e-06,
+      "loss": 1.0216,
+      "step": 7795
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6395684858922328e-06,
+      "loss": 1.0126,
+      "step": 7800
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6340949943104617e-06,
+      "loss": 0.9929,
+      "step": 7805
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6286288691420914e-06,
+      "loss": 1.0062,
+      "step": 7810
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6231701223499519e-06,
+      "loss": 0.9688,
+      "step": 7815
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6177187658807182e-06,
+      "loss": 0.9612,
+      "step": 7820
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6122748116648984e-06,
+      "loss": 0.9724,
+      "step": 7825
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6068382716167957e-06,
+      "loss": 1.0499,
+      "step": 7830
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.601409157634487e-06,
+      "loss": 0.97,
+      "step": 7835
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.595987481599801e-06,
+      "loss": 1.0185,
+      "step": 7840
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.5905732553782827e-06,
+      "loss": 0.9706,
+      "step": 7845
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.5851664908191788e-06,
+      "loss": 0.9821,
+      "step": 7850
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.5797671997554016e-06,
+      "loss": 1.0146,
+      "step": 7855
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.5743753940035067e-06,
+      "loss": 0.973,
+      "step": 7860
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.568991085363673e-06,
+      "loss": 1.017,
+      "step": 7865
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.5636142856196652e-06,
+      "loss": 1.0181,
+      "step": 7870
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.5582450065388205e-06,
+      "loss": 0.9986,
+      "step": 7875
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.5528832598720117e-06,
+      "loss": 0.9966,
+      "step": 7880
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.54752905735363e-06,
+      "loss": 1.0512,
+      "step": 7885
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.5421824107015565e-06,
+      "loss": 1.0491,
+      "step": 7890
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.5368433316171321e-06,
+      "loss": 1.0325,
+      "step": 7895
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.531511831785138e-06,
+      "loss": 0.961,
+      "step": 7900
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.5261879228737697e-06,
+      "loss": 0.9795,
+      "step": 7905
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.520871616534606e-06,
+      "loss": 0.9812,
+      "step": 7910
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.515562924402591e-06,
+      "loss": 0.9549,
+      "step": 7915
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.5102618580960022e-06,
+      "loss": 0.9947,
+      "step": 7920
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.5049684292164268e-06,
+      "loss": 0.9944,
+      "step": 7925
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.4996826493487422e-06,
+      "loss": 1.0354,
+      "step": 7930
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.4944045300610776e-06,
+      "loss": 1.0001,
+      "step": 7935
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.4891340829048068e-06,
+      "loss": 0.9898,
+      "step": 7940
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.4838713194145044e-06,
+      "loss": 0.969,
+      "step": 7945
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.4786162511079305e-06,
+      "loss": 0.9292,
+      "step": 7950
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.4733688894860087e-06,
+      "loss": 0.9704,
+      "step": 7955
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.4681292460327905e-06,
+      "loss": 1.013,
+      "step": 7960
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.462897332215441e-06,
+      "loss": 1.0124,
+      "step": 7965
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.4576731594842053e-06,
+      "loss": 1.0041,
+      "step": 7970
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.4524567392723848e-06,
+      "loss": 1.0264,
+      "step": 7975
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.4472480829963215e-06,
+      "loss": 0.976,
+      "step": 7980
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.4420472020553572e-06,
+      "loss": 0.9835,
+      "step": 7985
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.436854107831825e-06,
+      "loss": 0.9903,
+      "step": 7990
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.4316688116910104e-06,
+      "loss": 1.0028,
+      "step": 7995
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.426491324981134e-06,
+      "loss": 1.0626,
+      "step": 8000
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.4213216590333267e-06,
+      "loss": 0.9957,
+      "step": 8005
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.4161598251616043e-06,
+      "loss": 0.9839,
+      "step": 8010
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.4110058346628375e-06,
+      "loss": 1.0247,
+      "step": 8015
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.4058596988167361e-06,
+      "loss": 1.0008,
+      "step": 8020
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.4007214288858156e-06,
+      "loss": 1.0017,
+      "step": 8025
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.3955910361153806e-06,
+      "loss": 1.0104,
+      "step": 8030
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.3904685317334926e-06,
+      "loss": 0.9825,
+      "step": 8035
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.3853539269509525e-06,
+      "loss": 0.9759,
+      "step": 8040
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.38024723296127e-06,
+      "loss": 0.9844,
+      "step": 8045
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.3751484609406407e-06,
+      "loss": 1.0319,
+      "step": 8050
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.370057622047928e-06,
+      "loss": 1.0118,
+      "step": 8055
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.3649747274246267e-06,
+      "loss": 1.0164,
+      "step": 8060
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.3598997881948512e-06,
+      "loss": 1.0283,
+      "step": 8065
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.3548328154653013e-06,
+      "loss": 0.9834,
+      "step": 8070
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.3497738203252415e-06,
+      "loss": 0.9829,
+      "step": 8075
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.3447228138464812e-06,
+      "loss": 0.9738,
+      "step": 8080
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.3396798070833406e-06,
+      "loss": 0.9093,
+      "step": 8085
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.3346448110726385e-06,
+      "loss": 0.9741,
+      "step": 8090
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.329617836833657e-06,
+      "loss": 1.0285,
+      "step": 8095
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.3245988953681221e-06,
+      "loss": 1.067,
+      "step": 8100
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.319587997660185e-06,
+      "loss": 1.0426,
+      "step": 8105
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.3145851546763855e-06,
+      "loss": 0.9884,
+      "step": 8110
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.3095903773656432e-06,
+      "loss": 0.9974,
+      "step": 8115
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.304603676659219e-06,
+      "loss": 1.0406,
+      "step": 8120
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.299625063470698e-06,
+      "loss": 1.0303,
+      "step": 8125
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.2946545486959739e-06,
+      "loss": 1.0013,
+      "step": 8130
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.2896921432132076e-06,
+      "loss": 1.0501,
+      "step": 8135
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.2847378578828145e-06,
+      "loss": 1.035,
+      "step": 8140
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.2797917035474421e-06,
+      "loss": 0.9811,
+      "step": 8145
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.2748536910319381e-06,
+      "loss": 0.9942,
+      "step": 8150
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.2699238311433365e-06,
+      "loss": 0.9914,
+      "step": 8155
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.2650021346708258e-06,
+      "loss": 1.0016,
+      "step": 8160
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.2600886123857276e-06,
+      "loss": 1.0056,
+      "step": 8165
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.2551832750414782e-06,
+      "loss": 1.0331,
+      "step": 8170
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.2502861333735954e-06,
+      "loss": 0.9904,
+      "step": 8175
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.2453971980996665e-06,
+      "loss": 1.0247,
+      "step": 8180
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.2405164799193143e-06,
+      "loss": 1.0195,
+      "step": 8185
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.235643989514177e-06,
+      "loss": 1.0077,
+      "step": 8190
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.2307797375478925e-06,
+      "loss": 0.9906,
+      "step": 8195
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.2259237346660613e-06,
+      "loss": 0.9833,
+      "step": 8200
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.2210759914962367e-06,
+      "loss": 0.9782,
+      "step": 8205
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.2162365186478902e-06,
+      "loss": 1.0459,
+      "step": 8210
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.211405326712395e-06,
+      "loss": 0.9974,
+      "step": 8215
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.2065824262630037e-06,
+      "loss": 0.9904,
+      "step": 8220
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.2017678278548184e-06,
+      "loss": 0.9625,
+      "step": 8225
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.196961542024776e-06,
+      "loss": 0.9987,
+      "step": 8230
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.1921635792916185e-06,
+      "loss": 0.987,
+      "step": 8235
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.1873739501558723e-06,
+      "loss": 1.0117,
+      "step": 8240
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.1825926650998276e-06,
+      "loss": 1.0068,
+      "step": 8245
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.1778197345875092e-06,
+      "loss": 1.0296,
+      "step": 8250
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.1730551690646624e-06,
+      "loss": 0.9519,
+      "step": 8255
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.1682989789587246e-06,
+      "loss": 0.9607,
+      "step": 8260
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.1635511746787985e-06,
+      "loss": 0.9565,
+      "step": 8265
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.1588117666156417e-06,
+      "loss": 1.0062,
+      "step": 8270
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.1540807651416274e-06,
+      "loss": 0.9759,
+      "step": 8275
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.14935818061074e-06,
+      "loss": 0.9834,
+      "step": 8280
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.1446440233585365e-06,
+      "loss": 1.0031,
+      "step": 8285
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.1399383037021317e-06,
+      "loss": 1.0102,
+      "step": 8290
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.135241031940177e-06,
+      "loss": 1.006,
+      "step": 8295
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.130552218352831e-06,
+      "loss": 0.9901,
+      "step": 8300
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.125871873201747e-06,
+      "loss": 1.0177,
+      "step": 8305
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.1212000067300399e-06,
+      "loss": 1.0048,
+      "step": 8310
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.1165366291622687e-06,
+      "loss": 1.0037,
+      "step": 8315
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.1118817507044188e-06,
+      "loss": 0.9942,
+      "step": 8320
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.1072353815438691e-06,
+      "loss": 0.9472,
+      "step": 8325
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.1025975318493814e-06,
+      "loss": 1.0209,
+      "step": 8330
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.0979682117710671e-06,
+      "loss": 1.0373,
+      "step": 8335
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.093347431440372e-06,
+      "loss": 1.0145,
+      "step": 8340
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.0887352009700548e-06,
+      "loss": 1.0336,
+      "step": 8345
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.0841315304541578e-06,
+      "loss": 1.0059,
+      "step": 8350
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.079536429967995e-06,
+      "loss": 0.9977,
+      "step": 8355
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.0749499095681204e-06,
+      "loss": 0.9639,
+      "step": 8360
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.07037197929231e-06,
+      "loss": 1.0021,
+      "step": 8365
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.0658026491595442e-06,
+      "loss": 1.0374,
+      "step": 8370
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.061241929169976e-06,
+      "loss": 1.007,
+      "step": 8375
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.0566898293049188e-06,
+      "loss": 0.9839,
+      "step": 8380
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.0521463595268217e-06,
+      "loss": 0.9828,
+      "step": 8385
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.04761152977924e-06,
+      "loss": 0.9932,
+      "step": 8390
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.0430853499868282e-06,
+      "loss": 0.9747,
+      "step": 8395
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.0385678300553043e-06,
+      "loss": 1.0261,
+      "step": 8400
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.0340589798714334e-06,
+      "loss": 0.9775,
+      "step": 8405
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.0295588093030123e-06,
+      "loss": 1.0098,
+      "step": 8410
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.0250673281988355e-06,
+      "loss": 0.9396,
+      "step": 8415
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.0205845463886855e-06,
+      "loss": 1.0189,
+      "step": 8420
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.0161104736833033e-06,
+      "loss": 0.9858,
+      "step": 8425
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.0116451198743682e-06,
+      "loss": 1.0138,
+      "step": 8430
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.0071884947344835e-06,
+      "loss": 0.958,
+      "step": 8435
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.0027406080171421e-06,
+      "loss": 0.9922,
+      "step": 8440
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.983014694567189e-07,
+      "loss": 0.9678,
+      "step": 8445
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.9387108876844e-07,
+      "loss": 1.0036,
+      "step": 8450
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.89449475648363e-07,
+      "loss": 1.0105,
+      "step": 8455
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.850366397733624e-07,
+      "loss": 0.9776,
+      "step": 8460
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.806325908010978e-07,
+      "loss": 0.9267,
+      "step": 8465
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.76237338370003e-07,
+      "loss": 1.0401,
+      "step": 8470
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.718508920992576e-07,
+      "loss": 1.0059,
+      "step": 8475
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.674732615887666e-07,
+      "loss": 1.0046,
+      "step": 8480
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.631044564191473e-07,
+      "loss": 1.0045,
+      "step": 8485
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.587444861516964e-07,
+      "loss": 0.9767,
+      "step": 8490
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.543933603283794e-07,
+      "loss": 1.0321,
+      "step": 8495
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.500510884718017e-07,
+      "loss": 1.015,
+      "step": 8500
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.457176800851948e-07,
+      "loss": 0.9711,
+      "step": 8505
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.413931446523917e-07,
+      "loss": 1.0119,
+      "step": 8510
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.370774916378022e-07,
+      "loss": 1.0316,
+      "step": 8515
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.327707304864042e-07,
+      "loss": 0.9668,
+      "step": 8520
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.284728706237089e-07,
+      "loss": 0.9267,
+      "step": 8525
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.241839214557474e-07,
+      "loss": 1.0602,
+      "step": 8530
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.199038923690545e-07,
+      "loss": 0.983,
+      "step": 8535
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.156327927306363e-07,
+      "loss": 0.9854,
+      "step": 8540
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.113706318879623e-07,
+      "loss": 0.9946,
+      "step": 8545
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.071174191689358e-07,
+      "loss": 1.0014,
+      "step": 8550
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.028731638818756e-07,
+      "loss": 1.0559,
+      "step": 8555
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.986378753155012e-07,
+      "loss": 0.9842,
+      "step": 8560
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.944115627389032e-07,
+      "loss": 1.0126,
+      "step": 8565
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.90194235401533e-07,
+      "loss": 1.0187,
+      "step": 8570
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.859859025331735e-07,
+      "loss": 0.9868,
+      "step": 8575
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.81786573343923e-07,
+      "loss": 1.0083,
+      "step": 8580
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.775962570241775e-07,
+      "loss": 1.0306,
+      "step": 8585
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.734149627446048e-07,
+      "loss": 0.9807,
+      "step": 8590
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.692426996561315e-07,
+      "loss": 0.9988,
+      "step": 8595
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.650794768899162e-07,
+      "loss": 0.9973,
+      "step": 8600
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.609253035573311e-07,
+      "loss": 0.9622,
+      "step": 8605
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.567801887499483e-07,
+      "loss": 0.9839,
+      "step": 8610
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.526441415395093e-07,
+      "loss": 1.0232,
+      "step": 8615
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.485171709779161e-07,
+      "loss": 1.0403,
+      "step": 8620
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.443992860972011e-07,
+      "loss": 1.0501,
+      "step": 8625
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.40290495909516e-07,
+      "loss": 0.9862,
+      "step": 8630
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.36190809407108e-07,
+      "loss": 0.9605,
+      "step": 8635
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.321002355622993e-07,
+      "loss": 1.0393,
+      "step": 8640
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.280187833274672e-07,
+      "loss": 0.9765,
+      "step": 8645
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.239464616350307e-07,
+      "loss": 1.0238,
+      "step": 8650
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.19883279397421e-07,
+      "loss": 0.9585,
+      "step": 8655
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.158292455070721e-07,
+      "loss": 1.0253,
+      "step": 8660
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.117843688363925e-07,
+      "loss": 0.9903,
+      "step": 8665
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.077486582377508e-07,
+      "loss": 1.0217,
+      "step": 8670
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.037221225434583e-07,
+      "loss": 0.9856,
+      "step": 8675
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 7.997047705657412e-07,
+      "loss": 0.9672,
+      "step": 8680
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.956966110967324e-07,
+      "loss": 0.9625,
+      "step": 8685
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.91697652908443e-07,
+      "loss": 1.0186,
+      "step": 8690
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.877079047527469e-07,
+      "loss": 1.0295,
+      "step": 8695
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.837273753613645e-07,
+      "loss": 1.0459,
+      "step": 8700
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.797560734458359e-07,
+      "loss": 0.9633,
+      "step": 8705
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.757940076975117e-07,
+      "loss": 0.9896,
+      "step": 8710
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.718411867875253e-07,
+      "loss": 0.97,
+      "step": 8715
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.678976193667764e-07,
+      "loss": 0.9848,
+      "step": 8720
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.639633140659175e-07,
+      "loss": 0.9779,
+      "step": 8725
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.600382794953259e-07,
+      "loss": 0.9598,
+      "step": 8730
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.561225242450943e-07,
+      "loss": 1.0434,
+      "step": 8735
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.522160568850018e-07,
+      "loss": 1.0151,
+      "step": 8740
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.483188859645047e-07,
+      "loss": 0.9849,
+      "step": 8745
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.444310200127136e-07,
+      "loss": 0.9867,
+      "step": 8750
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.405524675383713e-07,
+      "loss": 1.0014,
+      "step": 8755
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.366832370298427e-07,
+      "loss": 1.0249,
+      "step": 8760
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.328233369550874e-07,
+      "loss": 0.968,
+      "step": 8765
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.289727757616444e-07,
+      "loss": 1.0253,
+      "step": 8770
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.251315618766186e-07,
+      "loss": 0.9783,
+      "step": 8775
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.212997037066527e-07,
+      "loss": 1.0391,
+      "step": 8780
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.17477209637919e-07,
+      "loss": 0.9626,
+      "step": 8785
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.136640880360935e-07,
+      "loss": 1.0033,
+      "step": 8790
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.098603472463384e-07,
+      "loss": 0.9924,
+      "step": 8795
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.060659955932897e-07,
+      "loss": 0.9939,
+      "step": 8800
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.022810413810311e-07,
+      "loss": 0.9911,
+      "step": 8805
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 6.985054928930835e-07,
+      "loss": 0.969,
+      "step": 8810
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 6.9473935839238e-07,
+      "loss": 0.9968,
+      "step": 8815
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 6.909826461212509e-07,
+      "loss": 0.9673,
+      "step": 8820
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 6.872353643014079e-07,
+      "loss": 0.9651,
+      "step": 8825
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 6.834975211339207e-07,
+      "loss": 0.982,
+      "step": 8830
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 6.797691247992055e-07,
+      "loss": 1.0256,
+      "step": 8835
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 6.760501834570011e-07,
+      "loss": 0.9187,
+      "step": 8840
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 6.723407052463538e-07,
+      "loss": 1.0243,
+      "step": 8845
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 6.68640698285602e-07,
+      "loss": 0.9892,
+      "step": 8850
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 6.649501706723522e-07,
+      "loss": 1.0034,
+      "step": 8855
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 6.612691304834695e-07,
+      "loss": 0.9674,
+      "step": 8860
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 6.57597585775051e-07,
+      "loss": 1.0126,
+      "step": 8865
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 6.539355445824125e-07,
+      "loss": 1.0072,
+      "step": 8870
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 6.502830149200773e-07,
+      "loss": 0.9848,
+      "step": 8875
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 6.466400047817461e-07,
+      "loss": 0.991,
+      "step": 8880
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 6.430065221402865e-07,
+      "loss": 0.9503,
+      "step": 8885
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 6.393825749477178e-07,
+      "loss": 0.9877,
+      "step": 8890
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.357681711351865e-07,
+      "loss": 0.9854,
+      "step": 8895
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.321633186129578e-07,
+      "loss": 1.0191,
+      "step": 8900
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.285680252703891e-07,
+      "loss": 1.0263,
+      "step": 8905
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.249822989759191e-07,
+      "loss": 0.9759,
+      "step": 8910
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.214061475770499e-07,
+      "loss": 0.9643,
+      "step": 8915
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.178395789003255e-07,
+      "loss": 0.9902,
+      "step": 8920
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.142826007513214e-07,
+      "loss": 1.0067,
+      "step": 8925
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.107352209146211e-07,
+      "loss": 1.0193,
+      "step": 8930
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.071974471538023e-07,
+      "loss": 0.9872,
+      "step": 8935
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.036692872114203e-07,
+      "loss": 0.9438,
+      "step": 8940
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.001507488089891e-07,
+      "loss": 0.9497,
+      "step": 8945
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 5.966418396469676e-07,
+      "loss": 0.9883,
+      "step": 8950
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 5.931425674047386e-07,
+      "loss": 1.0051,
+      "step": 8955
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 5.896529397405936e-07,
+      "loss": 0.955,
+      "step": 8960
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 5.861729642917202e-07,
+      "loss": 0.9636,
+      "step": 8965
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 5.827026486741766e-07,
+      "loss": 1.007,
+      "step": 8970
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 5.792420004828859e-07,
+      "loss": 0.9726,
+      "step": 8975
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 5.757910272916078e-07,
+      "loss": 0.9744,
+      "step": 8980
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 5.723497366529329e-07,
+      "loss": 1.0138,
+      "step": 8985
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 5.689181360982588e-07,
+      "loss": 1.0067,
+      "step": 8990
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 5.654962331377728e-07,
+      "loss": 0.9926,
+      "step": 8995
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.620840352604473e-07,
+      "loss": 1.0074,
+      "step": 9000
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.586815499340064e-07,
+      "loss": 0.9769,
+      "step": 9005
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.552887846049204e-07,
+      "loss": 0.9856,
+      "step": 9010
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.519057466983896e-07,
+      "loss": 1.044,
+      "step": 9015
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.485324436183214e-07,
+      "loss": 0.92,
+      "step": 9020
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.451688827473217e-07,
+      "loss": 0.9903,
+      "step": 9025
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.418150714466736e-07,
+      "loss": 1.021,
+      "step": 9030
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.384710170563212e-07,
+      "loss": 1.0316,
+      "step": 9035
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.351367268948593e-07,
+      "loss": 1.0387,
+      "step": 9040
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.318122082595089e-07,
+      "loss": 0.9826,
+      "step": 9045
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.284974684261101e-07,
+      "loss": 1.0169,
+      "step": 9050
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.251925146490977e-07,
+      "loss": 0.9853,
+      "step": 9055
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.218973541614902e-07,
+      "loss": 0.987,
+      "step": 9060
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.186119941748763e-07,
+      "loss": 0.962,
+      "step": 9065
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.15336441879391e-07,
+      "loss": 1.0041,
+      "step": 9070
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.120707044437095e-07,
+      "loss": 0.9375,
+      "step": 9075
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.088147890150236e-07,
+      "loss": 0.9796,
+      "step": 9080
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.055687027190293e-07,
+      "loss": 0.9812,
+      "step": 9085
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.023324526599139e-07,
+      "loss": 0.9641,
+      "step": 9090
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.991060459203339e-07,
+      "loss": 0.9975,
+      "step": 9095
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.95889489561408e-07,
+      "loss": 1.0031,
+      "step": 9100
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.926827906226928e-07,
+      "loss": 0.9925,
+      "step": 9105
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.894859561221715e-07,
+      "loss": 1.0428,
+      "step": 9110
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.862989930562434e-07,
+      "loss": 0.9984,
+      "step": 9115
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.83121908399698e-07,
+      "loss": 0.9843,
+      "step": 9120
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.799547091057083e-07,
+      "loss": 0.9559,
+      "step": 9125
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.767974021058147e-07,
+      "loss": 0.9861,
+      "step": 9130
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.736499943099032e-07,
+      "loss": 1.0061,
+      "step": 9135
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.7051249260619933e-07,
+      "loss": 1.0023,
+      "step": 9140
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.673849038612471e-07,
+      "loss": 0.9917,
+      "step": 9145
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.6426723491989344e-07,
+      "loss": 0.9885,
+      "step": 9150
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.6115949260527984e-07,
+      "loss": 0.9859,
+      "step": 9155
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.58061683718819e-07,
+      "loss": 0.999,
+      "step": 9160
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.5497381504018724e-07,
+      "loss": 0.976,
+      "step": 9165
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.518958933273038e-07,
+      "loss": 0.99,
+      "step": 9170
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.488279253163186e-07,
+      "loss": 1.0054,
+      "step": 9175
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.4576991772160014e-07,
+      "loss": 0.9667,
+      "step": 9180
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.4272187723571446e-07,
+      "loss": 1.0176,
+      "step": 9185
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.3968381052941876e-07,
+      "loss": 0.9915,
+      "step": 9190
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.366557242516378e-07,
+      "loss": 1.0613,
+      "step": 9195
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.3363762502945496e-07,
+      "loss": 1.0286,
+      "step": 9200
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.306295194680987e-07,
+      "loss": 0.997,
+      "step": 9205
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.2763141415092414e-07,
+      "loss": 0.972,
+      "step": 9210
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.2464331563940085e-07,
+      "loss": 1.0305,
+      "step": 9215
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.216652304730973e-07,
+      "loss": 1.055,
+      "step": 9220
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.1869716516966975e-07,
+      "loss": 1.0262,
+      "step": 9225
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.157391262248428e-07,
+      "loss": 1.0102,
+      "step": 9230
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.1279112011239854e-07,
+      "loss": 0.9798,
+      "step": 9235
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.098531532841643e-07,
+      "loss": 1.0373,
+      "step": 9240
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.069252321699929e-07,
+      "loss": 1.0202,
+      "step": 9245
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.0400736317775303e-07,
+      "loss": 0.9893,
+      "step": 9250
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.0109955269331534e-07,
+      "loss": 0.9882,
+      "step": 9255
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.982018070805338e-07,
+      "loss": 1.014,
+      "step": 9260
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.953141326812382e-07,
+      "loss": 0.9922,
+      "step": 9265
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.9243653581521544e-07,
+      "loss": 0.9957,
+      "step": 9270
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.895690227801968e-07,
+      "loss": 0.9637,
+      "step": 9275
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.8671159985184725e-07,
+      "loss": 0.9959,
+      "step": 9280
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.8386427328374496e-07,
+      "loss": 1.0288,
+      "step": 9285
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.81027049307377e-07,
+      "loss": 0.9789,
+      "step": 9290
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.781999341321169e-07,
+      "loss": 1.0493,
+      "step": 9295
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.753829339452147e-07,
+      "loss": 1.0578,
+      "step": 9300
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.725760549117857e-07,
+      "loss": 0.9472,
+      "step": 9305
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.6977930317479206e-07,
+      "loss": 0.9805,
+      "step": 9310
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.6699268485503426e-07,
+      "loss": 0.9856,
+      "step": 9315
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.642162060511345e-07,
+      "loss": 1.024,
+      "step": 9320
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.614498728395227e-07,
+      "loss": 0.9993,
+      "step": 9325
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.5869369127442823e-07,
+      "loss": 0.9743,
+      "step": 9330
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.5594766738785926e-07,
+      "loss": 0.987,
+      "step": 9335
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.5321180718959627e-07,
+      "loss": 0.9956,
+      "step": 9340
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.5048611666717583e-07,
+      "loss": 0.996,
+      "step": 9345
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.4777060178587506e-07,
+      "loss": 0.9994,
+      "step": 9350
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.450652684887057e-07,
+      "loss": 1.0151,
+      "step": 9355
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.4237012269639325e-07,
+      "loss": 0.9988,
+      "step": 9360
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.3968517030736893e-07,
+      "loss": 0.9518,
+      "step": 9365
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.370104171977545e-07,
+      "loss": 1.0358,
+      "step": 9370
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.343458692213508e-07,
+      "loss": 0.9904,
+      "step": 9375
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.316915322096248e-07,
+      "loss": 0.9831,
+      "step": 9380
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.290474119716963e-07,
+      "loss": 0.9646,
+      "step": 9385
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.2641351429432257e-07,
+      "loss": 0.9685,
+      "step": 9390
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.2378984494189305e-07,
+      "loss": 1.0001,
+      "step": 9395
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.2117640965640794e-07,
+      "loss": 0.9918,
+      "step": 9400
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.185732141574732e-07,
+      "loss": 0.9764,
+      "step": 9405
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.159802641422816e-07,
+      "loss": 1.0041,
+      "step": 9410
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.1339756528560473e-07,
+      "loss": 0.9795,
+      "step": 9415
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.1082512323977965e-07,
+      "loss": 0.9923,
+      "step": 9420
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.0826294363469456e-07,
+      "loss": 0.9992,
+      "step": 9425
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.0571103207778006e-07,
+      "loss": 1.0075,
+      "step": 9430
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.03169394153992e-07,
+      "loss": 1.0872,
+      "step": 9435
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.006380354258037e-07,
+      "loss": 1.0026,
+      "step": 9440
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.9811696143319246e-07,
+      "loss": 0.9731,
+      "step": 9445
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.956061776936248e-07,
+      "loss": 1.0383,
+      "step": 9450
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.931056897020501e-07,
+      "loss": 0.9793,
+      "step": 9455
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.9061550293088104e-07,
+      "loss": 0.9794,
+      "step": 9460
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.8813562282998973e-07,
+      "loss": 0.9626,
+      "step": 9465
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.856660548266887e-07,
+      "loss": 0.9648,
+      "step": 9470
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.8320680432572245e-07,
+      "loss": 0.9974,
+      "step": 9475
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.807578767092578e-07,
+      "loss": 0.9994,
+      "step": 9480
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.7831927733686595e-07,
+      "loss": 0.9218,
+      "step": 9485
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.7589101154551657e-07,
+      "loss": 0.9894,
+      "step": 9490
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.7347308464956367e-07,
+      "loss": 1.0228,
+      "step": 9495
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.710655019407332e-07,
+      "loss": 0.9644,
+      "step": 9500
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.686682686881148e-07,
+      "loss": 0.9731,
+      "step": 9505
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.662813901381445e-07,
+      "loss": 1.0288,
+      "step": 9510
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.639048715145981e-07,
+      "loss": 0.9898,
+      "step": 9515
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.615387180185802e-07,
+      "loss": 0.9612,
+      "step": 9520
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.591829348285074e-07,
+      "loss": 0.9785,
+      "step": 9525
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.56837527100105e-07,
+      "loss": 0.9774,
+      "step": 9530
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.545024999663864e-07,
+      "loss": 0.9833,
+      "step": 9535
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.5217785853764944e-07,
+      "loss": 1.0435,
+      "step": 9540
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.498636079014627e-07,
+      "loss": 1.0122,
+      "step": 9545
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.47559753122652e-07,
+      "loss": 0.9966,
+      "step": 9550
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.452662992432936e-07,
+      "loss": 1.0294,
+      "step": 9555
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.4298325128270017e-07,
+      "loss": 1.0121,
+      "step": 9560
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.407106142374094e-07,
+      "loss": 0.9641,
+      "step": 9565
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.3844839308117653e-07,
+      "loss": 0.9893,
+      "step": 9570
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.3619659276495954e-07,
+      "loss": 0.998,
+      "step": 9575
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.339552182169119e-07,
+      "loss": 0.9964,
+      "step": 9580
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.31724274342367e-07,
+      "loss": 1.0028,
+      "step": 9585
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.2950376602383195e-07,
+      "loss": 0.9907,
+      "step": 9590
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.2729369812097667e-07,
+      "loss": 1.02,
+      "step": 9595
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.2509407547061823e-07,
+      "loss": 0.9512,
+      "step": 9600
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.2290490288671807e-07,
+      "loss": 0.9573,
+      "step": 9605
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.2072618516036372e-07,
+      "loss": 0.9834,
+      "step": 9610
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.1855792705976264e-07,
+      "loss": 1.0152,
+      "step": 9615
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.1640013333023402e-07,
+      "loss": 0.9628,
+      "step": 9620
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.1425280869419085e-07,
+      "loss": 1.0288,
+      "step": 9625
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.1211595785113727e-07,
+      "loss": 0.9941,
+      "step": 9630
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.0998958547765347e-07,
+      "loss": 0.9589,
+      "step": 9635
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.0787369622738807e-07,
+      "loss": 0.9729,
+      "step": 9640
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.0576829473104687e-07,
+      "loss": 0.9687,
+      "step": 9645
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.0367338559638184e-07,
+      "loss": 0.9604,
+      "step": 9650
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.0158897340818274e-07,
+      "loss": 0.9401,
+      "step": 9655
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9951506272826716e-07,
+      "loss": 0.9812,
+      "step": 9660
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9745165809546772e-07,
+      "loss": 0.9752,
+      "step": 9665
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9539876402562717e-07,
+      "loss": 0.9794,
+      "step": 9670
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9335638501158216e-07,
+      "loss": 1.0045,
+      "step": 9675
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9132452552315772e-07,
+      "loss": 1.0158,
+      "step": 9680
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.8930319000715847e-07,
+      "loss": 0.9873,
+      "step": 9685
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.8729238288735406e-07,
+      "loss": 0.944,
+      "step": 9690
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.8529210856447545e-07,
+      "loss": 0.9513,
+      "step": 9695
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.8330237141619912e-07,
+      "loss": 1.0099,
+      "step": 9700
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.8132317579714232e-07,
+      "loss": 0.9773,
+      "step": 9705
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.7935452603885074e-07,
+      "loss": 0.9655,
+      "step": 9710
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.7739642644979071e-07,
+      "loss": 0.9862,
+      "step": 9715
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.7544888131533987e-07,
+      "loss": 1.0405,
+      "step": 9720
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.7351189489777486e-07,
+      "loss": 1.0087,
+      "step": 9725
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.715854714362669e-07,
+      "loss": 1.0246,
+      "step": 9730
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.6966961514686743e-07,
+      "loss": 0.9786,
+      "step": 9735
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.677643302225007e-07,
+      "loss": 0.9962,
+      "step": 9740
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.658696208329591e-07,
+      "loss": 0.9492,
+      "step": 9745
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.639854911248867e-07,
+      "loss": 1.0017,
+      "step": 9750
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6211194522177287e-07,
+      "loss": 0.9738,
+      "step": 9755
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6024898722394765e-07,
+      "loss": 0.9666,
+      "step": 9760
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.583966212085647e-07,
+      "loss": 0.9243,
+      "step": 9765
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.565548512296e-07,
+      "loss": 1.0617,
+      "step": 9770
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.5472368131783766e-07,
+      "loss": 1.0325,
+      "step": 9775
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.5290311548086412e-07,
+      "loss": 1.0123,
+      "step": 9780
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.5109315770305942e-07,
+      "loss": 0.9989,
+      "step": 9785
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.492938119455839e-07,
+      "loss": 0.9708,
+      "step": 9790
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.475050821463775e-07,
+      "loss": 1.0057,
+      "step": 9795
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.457269722201432e-07,
+      "loss": 0.9792,
+      "step": 9800
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.439594860583443e-07,
+      "loss": 0.9827,
+      "step": 9805
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.422026275291921e-07,
+      "loss": 1.0001,
+      "step": 9810
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4045640047763986e-07,
+      "loss": 1.008,
+      "step": 9815
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.387208087253733e-07,
+      "loss": 0.9918,
+      "step": 9820
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.369958560708018e-07,
+      "loss": 0.9983,
+      "step": 9825
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.3528154628905055e-07,
+      "loss": 1.0213,
+      "step": 9830
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.3357788313195397e-07,
+      "loss": 1.0056,
+      "step": 9835
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.3188487032804397e-07,
+      "loss": 1.0007,
+      "step": 9840
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.3020251158254505e-07,
+      "loss": 0.971,
+      "step": 9845
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.2853081057736415e-07,
+      "loss": 0.9911,
+      "step": 9850
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.2686977097108366e-07,
+      "loss": 0.9954,
+      "step": 9855
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.2521939639895343e-07,
+      "loss": 1.0268,
+      "step": 9860
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.2357969047288143e-07,
+      "loss": 0.9966,
+      "step": 9865
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.2195065678142826e-07,
+      "loss": 1.0024,
+      "step": 9870
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.2033229888979704e-07,
+      "loss": 1.0428,
+      "step": 9875
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.1872462033982624e-07,
+      "loss": 0.9671,
+      "step": 9880
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.1712762464998307e-07,
+      "loss": 0.9602,
+      "step": 9885
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.155413153153545e-07,
+      "loss": 0.9832,
+      "step": 9890
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.1396569580763905e-07,
+      "loss": 0.9698,
+      "step": 9895
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.1240076957514224e-07,
+      "loss": 0.9876,
+      "step": 9900
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.108465400427644e-07,
+      "loss": 0.9923,
+      "step": 9905
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.0930301061199856e-07,
+      "loss": 0.9866,
+      "step": 9910
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.0777018466091749e-07,
+      "loss": 0.9829,
+      "step": 9915
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.0624806554417e-07,
+      "loss": 1.0067,
+      "step": 9920
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.0473665659297361e-07,
+      "loss": 1.0359,
+      "step": 9925
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.032359611151046e-07,
+      "loss": 0.9923,
+      "step": 9930
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.0174598239489353e-07,
+      "loss": 0.9796,
+      "step": 9935
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.0026672369321589e-07,
+      "loss": 0.9877,
+      "step": 9940
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.879818824748755e-08,
+      "loss": 0.9713,
+      "step": 9945
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.734037927165486e-08,
+      "loss": 0.9924,
+      "step": 9950
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.589329995618845e-08,
+      "loss": 0.9724,
+      "step": 9955
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.445695346807837e-08,
+      "loss": 1.0025,
+      "step": 9960
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.303134295082339e-08,
+      "loss": 0.9767,
+      "step": 9965
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.161647152442887e-08,
+      "loss": 1.0034,
+      "step": 9970
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.021234228539511e-08,
+      "loss": 0.96,
+      "step": 9975
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 8.881895830671227e-08,
+      "loss": 1.0509,
+      "step": 9980
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 8.743632263785606e-08,
+      "loss": 0.9745,
+      "step": 9985
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 8.606443830477817e-08,
+      "loss": 0.9531,
+      "step": 9990
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 8.470330830990026e-08,
+      "loss": 1.0242,
+      "step": 9995
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 8.335293563210889e-08,
+      "loss": 0.995,
+      "step": 10000
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 8.201332322674504e-08,
+      "loss": 1.0034,
+      "step": 10005
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 8.068447402560464e-08,
+      "loss": 1.0096,
+      "step": 10010
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.93663909369241e-08,
+      "loss": 0.9714,
+      "step": 10015
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.805907684537984e-08,
+      "loss": 1.0069,
+      "step": 10020
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.676253461207873e-08,
+      "loss": 0.9708,
+      "step": 10025
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.547676707455431e-08,
+      "loss": 0.9917,
+      "step": 10030
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.420177704675846e-08,
+      "loss": 1.0077,
+      "step": 10035
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.293756731905633e-08,
+      "loss": 0.9175,
+      "step": 10040
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.168414065821917e-08,
+      "loss": 1.0149,
+      "step": 10045
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.044149980742099e-08,
+      "loss": 0.9853,
+      "step": 10050
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.920964748622805e-08,
+      "loss": 1.0303,
+      "step": 10055
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.798858639059768e-08,
+      "loss": 1.0283,
+      "step": 10060
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.677831919286892e-08,
+      "loss": 0.9688,
+      "step": 10065
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.557884854175855e-08,
+      "loss": 0.994,
+      "step": 10070
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.43901770623545e-08,
+      "loss": 0.9554,
+      "step": 10075
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.321230735611083e-08,
+      "loss": 1.0381,
+      "step": 10080
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.204524200083994e-08,
+      "loss": 0.9653,
+      "step": 10085
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.088898355071094e-08,
+      "loss": 0.9959,
+      "step": 10090
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.974353453623849e-08,
+      "loss": 1.0315,
+      "step": 10095
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.8608897464283444e-08,
+      "loss": 0.9617,
+      "step": 10100
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.748507481804222e-08,
+      "loss": 1.0278,
+      "step": 10105
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.637206905704462e-08,
+      "loss": 1.0349,
+      "step": 10110
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.526988261714661e-08,
+      "loss": 0.9845,
+      "step": 10115
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.417851791052642e-08,
+      "loss": 0.9364,
+      "step": 10120
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.309797732567734e-08,
+      "loss": 1.0102,
+      "step": 10125
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.2028263227404393e-08,
+      "loss": 1.0084,
+      "step": 10130
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.0969377956817664e-08,
+      "loss": 0.9695,
+      "step": 10135
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.992132383133008e-08,
+      "loss": 0.9675,
+      "step": 10140
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.888410314464742e-08,
+      "loss": 0.9477,
+      "step": 10145
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.7857718166767764e-08,
+      "loss": 1.0022,
+      "step": 10150
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.6842171143974824e-08,
+      "loss": 1.0033,
+      "step": 10155
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.583746429883129e-08,
+      "loss": 1.0093,
+      "step": 10160
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.484359983017828e-08,
+      "loss": 1.0255,
+      "step": 10165
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.386057991312531e-08,
+      "loss": 1.0021,
+      "step": 10170
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.288840669905148e-08,
+      "loss": 1.0019,
+      "step": 10175
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.1927082315593725e-08,
+      "loss": 0.9963,
+      "step": 10180
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.0976608866648005e-08,
+      "loss": 1.0084,
+      "step": 10185
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.003698843236314e-08,
+      "loss": 0.9378,
+      "step": 10190
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.91082230691342e-08,
+      "loss": 0.9991,
+      "step": 10195
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.8190314809600226e-08,
+      "loss": 0.981,
+      "step": 10200
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.728326566263818e-08,
+      "loss": 1.0188,
+      "step": 10205
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.63870776133618e-08,
+      "loss": 0.9712,
+      "step": 10210
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.5501752623112174e-08,
+      "loss": 1.0227,
+      "step": 10215
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.462729262945774e-08,
+      "loss": 1.0069,
+      "step": 10220
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.376369954618819e-08,
+      "loss": 1.0695,
+      "step": 10225
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.291097526330944e-08,
+      "loss": 0.9937,
+      "step": 10230
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.2069121647043125e-08,
+      "loss": 1.0103,
+      "step": 10235
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.1238140539817105e-08,
+      "loss": 0.972,
+      "step": 10240
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.041803376026664e-08,
+      "loss": 0.9385,
+      "step": 10245
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.960880310322656e-08,
+      "loss": 1.0596,
+      "step": 10250
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.8810450339730177e-08,
+      "loss": 0.9672,
+      "step": 10255
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.8022977217002646e-08,
+      "loss": 0.9674,
+      "step": 10260
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.7246385458459814e-08,
+      "loss": 0.9845,
+      "step": 10265
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.6480676763702696e-08,
+      "loss": 1.0514,
+      "step": 10270
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.572585280851414e-08,
+      "loss": 0.9756,
+      "step": 10275
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.4981915244856047e-08,
+      "loss": 0.9719,
+      "step": 10280
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.4248865700864377e-08,
+      "loss": 0.9828,
+      "step": 10285
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.3526705780846926e-08,
+      "loss": 1.0349,
+      "step": 10290
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.2815437065277223e-08,
+      "loss": 1.0193,
+      "step": 10295
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.21150611107962e-08,
+      "loss": 0.9598,
+      "step": 10300
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.1425579450202185e-08,
+      "loss": 0.9831,
+      "step": 10305
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.074699359245369e-08,
+      "loss": 1.0146,
+      "step": 10310
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.0079305022661088e-08,
+      "loss": 1.0242,
+      "step": 10315
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.942251520208771e-08,
+      "loss": 0.9737,
+      "step": 10320
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.8776625568142637e-08,
+      "loss": 1.0052,
+      "step": 10325
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.8141637534380697e-08,
+      "loss": 0.9711,
+      "step": 10330
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7517552490498024e-08,
+      "loss": 0.982,
+      "step": 10335
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.690437180232818e-08,
+      "loss": 1.0527,
+      "step": 10340
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.6302096811841584e-08,
+      "loss": 1.0409,
+      "step": 10345
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.5710728837139976e-08,
+      "loss": 0.9326,
+      "step": 10350
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.5130269172455303e-08,
+      "loss": 0.9897,
+      "step": 10355
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.4560719088145825e-08,
+      "loss": 0.9835,
+      "step": 10360
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.4002079830693904e-08,
+      "loss": 1.0158,
+      "step": 10365
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3454352622703227e-08,
+      "loss": 0.9689,
+      "step": 10370
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.2917538662896023e-08,
+      "loss": 1.0638,
+      "step": 10375
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.23916391261103e-08,
+      "loss": 0.9282,
+      "step": 10380
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.1876655163298167e-08,
+      "loss": 1.026,
+      "step": 10385
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.1372587901521403e-08,
+      "loss": 1.0539,
+      "step": 10390
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0879438443950341e-08,
+      "loss": 0.9515,
+      "step": 10395
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0397207869862203e-08,
+      "loss": 0.9298,
+      "step": 10400
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.92589723463777e-09,
+      "loss": 1.0442,
+      "step": 10405
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.465507569757505e-09,
+      "loss": 1.0101,
+      "step": 10410
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.016039882802641e-09,
+      "loss": 0.9829,
+      "step": 10415
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 8.57749515745021e-09,
+      "loss": 1.0031,
+      "step": 10420
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 8.149874353473031e-09,
+      "loss": 0.9802,
+      "step": 10425
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.733178406735265e-09,
+      "loss": 0.9793,
+      "step": 10430
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.327408229191868e-09,
+      "loss": 0.9753,
+      "step": 10435
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.932564708887479e-09,
+      "loss": 1.0311,
+      "step": 10440
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.548648709951422e-09,
+      "loss": 0.9717,
+      "step": 10445
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.1756610725993706e-09,
+      "loss": 0.9608,
+      "step": 10450
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.8136026131294655e-09,
+      "loss": 0.984,
+      "step": 10455
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.462474123919536e-09,
+      "loss": 1.0421,
+      "step": 10460
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.122276373428769e-09,
+      "loss": 1.0026,
+      "step": 10465
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.793010106193264e-09,
+      "loss": 0.9449,
+      "step": 10470
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.474676042823811e-09,
+      "loss": 0.9989,
+      "step": 10475
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.167274880008676e-09,
+      "loss": 1.0245,
+      "step": 10480
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.870807290505263e-09,
+      "loss": 0.9744,
+      "step": 10485
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.585273923145671e-09,
+      "loss": 0.9621,
+      "step": 10490
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.310675402830588e-09,
+      "loss": 0.99,
+      "step": 10495
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.0470123305287314e-09,
+      "loss": 1.0376,
+      "step": 10500
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.7942852832779645e-09,
+      "loss": 1.0177,
+      "step": 10505
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.5524948141802952e-09,
+      "loss": 1.0608,
+      "step": 10510
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.3216414524041e-09,
+      "loss": 1.013,
+      "step": 10515
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.10172570318079e-09,
+      "loss": 1.0446,
+      "step": 10520
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.8927480478042605e-09,
+      "loss": 0.9628,
+      "step": 10525
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.694708943630885e-09,
+      "loss": 0.9825,
+      "step": 10530
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.5076088240756347e-09,
+      "loss": 1.0036,
+      "step": 10535
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3314480986159616e-09,
+      "loss": 0.9861,
+      "step": 10540
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.1662271527851377e-09,
+      "loss": 1.0007,
+      "step": 10545
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0119463481761405e-09,
+      "loss": 0.9734,
+      "step": 10550
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 8.68606022438323e-10,
+      "loss": 1.0294,
+      "step": 10555
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 7.362064892774134e-10,
+      "loss": 0.9354,
+      "step": 10560
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.147480384555149e-10,
+      "loss": 0.9794,
+      "step": 10565
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.042309357888853e-10,
+      "loss": 1.0016,
+      "step": 10570
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.0465542314793715e-10,
+      "loss": 0.9636,
+      "step": 10575
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.1602171845834805e-10,
+      "loss": 0.9528,
+      "step": 10580
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.3833001569772975e-10,
+      "loss": 0.9932,
+      "step": 10585
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.7158048489784863e-10,
+      "loss": 0.9984,
+      "step": 10590
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.1577327214240542e-10,
+      "loss": 0.9615,
+      "step": 10595
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 7.090849956759016e-11,
+      "loss": 1.0146,
+      "step": 10600
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.6986265362082276e-11,
+      "loss": 0.9899,
+      "step": 10605
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.4006643765385186e-11,
+      "loss": 1.0174,
+      "step": 10610
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9696850700468007e-12,
+      "loss": 1.0443,
+      "step": 10615
+    },
+    {
+      "epoch": 2.0,
+      "step": 10618,
+      "total_flos": 4.810450444432179e+17,
+      "train_loss": 1.1182698494837005,
+      "train_runtime": 99534.9844,
+      "train_samples_per_second": 13.655,
+      "train_steps_per_second": 0.107
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 10618,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "save_steps": 1000,
+  "total_flos": 4.810450444432179e+17,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}