diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,4819 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.37139342169401823,
+  "eval_steps": 500,
+  "global_step": 4000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 5e-06,
+      "loss": 2.6699,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1e-05,
+      "loss": 2.6758,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5e-05,
+      "loss": 2.6243,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2e-05,
+      "loss": 2.5668,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5e-05,
+      "loss": 2.5445,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3e-05,
+      "loss": 2.4871,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.5e-05,
+      "loss": 2.477,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4e-05,
+      "loss": 2.3967,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.5e-05,
+      "loss": 2.3669,
+      "step": 45
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5e-05,
+      "loss": 2.3135,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.500000000000001e-05,
+      "loss": 2.2882,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6e-05,
+      "loss": 2.2557,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.500000000000001e-05,
+      "loss": 2.2381,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7e-05,
+      "loss": 2.1791,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 2.1392,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8e-05,
+      "loss": 2.1101,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.5e-05,
+      "loss": 2.0965,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9e-05,
+      "loss": 2.1213,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.5e-05,
+      "loss": 2.0689,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001,
+      "loss": 2.0068,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99999458185223e-05,
+      "loss": 2.0107,
+      "step": 105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999978327420663e-05,
+      "loss": 2.0693,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999951236740525e-05,
+      "loss": 2.0156,
+      "step": 115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99991330987053e-05,
+      "loss": 1.9915,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999864546892874e-05,
+      "loss": 2.0002,
+      "step": 125
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999804947913241e-05,
+      "loss": 1.9707,
+      "step": 130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999734513060794e-05,
+      "loss": 1.9546,
+      "step": 135
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999653242488188e-05,
+      "loss": 1.9531,
+      "step": 140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999561136371555e-05,
+      "loss": 1.9414,
+      "step": 145
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999458194910512e-05,
+      "loss": 1.884,
+      "step": 150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999344418328162e-05,
+      "loss": 1.9035,
+      "step": 155
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999219806871085e-05,
+      "loss": 1.9181,
+      "step": 160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.999084360809351e-05,
+      "loss": 1.8555,
+      "step": 165
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.998938080436503e-05,
+      "loss": 1.848,
+      "step": 170
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99878096606957e-05,
+      "loss": 1.8488,
+      "step": 175
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.998613018049059e-05,
+      "loss": 1.848,
+      "step": 180
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.998434236738958e-05,
+      "loss": 1.8881,
+      "step": 185
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99824462252673e-05,
+      "loss": 1.8389,
+      "step": 190
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99804417582332e-05,
+      "loss": 1.7991,
+      "step": 195
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.997832897063148e-05,
+      "loss": 1.8291,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99761078670411e-05,
+      "loss": 1.8083,
+      "step": 205
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.997377845227576e-05,
+      "loss": 1.744,
+      "step": 210
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.997134073138389e-05,
+      "loss": 1.821,
+      "step": 215
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.996879470964868e-05,
+      "loss": 1.7805,
+      "step": 220
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.996614039258804e-05,
+      "loss": 1.8008,
+      "step": 225
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.996337778595453e-05,
+      "loss": 1.7867,
+      "step": 230
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.996050689573544e-05,
+      "loss": 1.7638,
+      "step": 235
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.995752772815274e-05,
+      "loss": 1.7697,
+      "step": 240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.995444028966307e-05,
+      "loss": 1.7381,
+      "step": 245
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.995124458695768e-05,
+      "loss": 1.7623,
+      "step": 250
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99479406269625e-05,
+      "loss": 1.7599,
+      "step": 255
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994452841683808e-05,
+      "loss": 1.7012,
+      "step": 260
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994100796397954e-05,
+      "loss": 1.738,
+      "step": 265
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.993737927601663e-05,
+      "loss": 1.7055,
+      "step": 270
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.993364236081366e-05,
+      "loss": 1.716,
+      "step": 275
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.992979722646948e-05,
+      "loss": 1.6933,
+      "step": 280
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.992584388131749e-05,
+      "loss": 1.68,
+      "step": 285
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.992178233392564e-05,
+      "loss": 1.6998,
+      "step": 290
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.991761259309634e-05,
+      "loss": 1.6955,
+      "step": 295
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.991333466786648e-05,
+      "loss": 1.6595,
+      "step": 300
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.990894856750744e-05,
+      "loss": 1.6446,
+      "step": 305
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.990445430152507e-05,
+      "loss": 1.7006,
+      "step": 310
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.989985187965956e-05,
+      "loss": 1.6734,
+      "step": 315
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.989514131188559e-05,
+      "loss": 1.6871,
+      "step": 320
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.989032260841216e-05,
+      "loss": 1.6623,
+      "step": 325
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.988539577968265e-05,
+      "loss": 1.6963,
+      "step": 330
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.988036083637479e-05,
+      "loss": 1.6186,
+      "step": 335
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.987521778940058e-05,
+      "loss": 1.6351,
+      "step": 340
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.986996664990636e-05,
+      "loss": 1.6056,
+      "step": 345
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.986460742927271e-05,
+      "loss": 1.6323,
+      "step": 350
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.985914013911443e-05,
+      "loss": 1.6609,
+      "step": 355
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.985356479128056e-05,
+      "loss": 1.6506,
+      "step": 360
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984788139785433e-05,
+      "loss": 1.6261,
+      "step": 365
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984208997115312e-05,
+      "loss": 1.5908,
+      "step": 370
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983619052372848e-05,
+      "loss": 1.6271,
+      "step": 375
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.9830183068366e-05,
+      "loss": 1.6192,
+      "step": 380
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.98240676180854e-05,
+      "loss": 1.558,
+      "step": 385
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.981784418614048e-05,
+      "loss": 1.6127,
+      "step": 390
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.981151278601899e-05,
+      "loss": 1.5966,
+      "step": 395
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.980507343144273e-05,
+      "loss": 1.6533,
+      "step": 400
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.979852613636745e-05,
+      "loss": 1.5718,
+      "step": 405
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.979187091498284e-05,
+      "loss": 1.5732,
+      "step": 410
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.978510778171248e-05,
+      "loss": 1.5729,
+      "step": 415
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.977823675121383e-05,
+      "loss": 1.5977,
+      "step": 420
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.97712578383782e-05,
+      "loss": 1.5877,
+      "step": 425
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.97641710583307e-05,
+      "loss": 1.5842,
+      "step": 430
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.975697642643023e-05,
+      "loss": 1.5584,
+      "step": 435
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.974967395826941e-05,
+      "loss": 1.5482,
+      "step": 440
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.974226366967458e-05,
+      "loss": 1.5785,
+      "step": 445
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.973474557670575e-05,
+      "loss": 1.559,
+      "step": 450
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.972711969565659e-05,
+      "loss": 1.5618,
+      "step": 455
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.971938604305435e-05,
+      "loss": 1.5639,
+      "step": 460
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.971154463565986e-05,
+      "loss": 1.5759,
+      "step": 465
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.970359549046749e-05,
+      "loss": 1.5296,
+      "step": 470
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.969553862470509e-05,
+      "loss": 1.5349,
+      "step": 475
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.968737405583396e-05,
+      "loss": 1.5089,
+      "step": 480
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.967910180154889e-05,
+      "loss": 1.5783,
+      "step": 485
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.967072187977795e-05,
+      "loss": 1.5102,
+      "step": 490
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.96622343086826e-05,
+      "loss": 1.5318,
+      "step": 495
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.965363910665761e-05,
+      "loss": 1.5893,
+      "step": 500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.964493629233104e-05,
+      "loss": 1.5346,
+      "step": 505
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.963612588456412e-05,
+      "loss": 1.5352,
+      "step": 510
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.962720790245128e-05,
+      "loss": 1.5867,
+      "step": 515
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.961818236532012e-05,
+      "loss": 1.5401,
+      "step": 520
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.960904929273129e-05,
+      "loss": 1.5437,
+      "step": 525
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.959980870447854e-05,
+      "loss": 1.5617,
+      "step": 530
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.959046062058863e-05,
+      "loss": 1.5396,
+      "step": 535
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.958100506132127e-05,
+      "loss": 1.5221,
+      "step": 540
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.957144204716908e-05,
+      "loss": 1.5521,
+      "step": 545
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.956177159885765e-05,
+      "loss": 1.5133,
+      "step": 550
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.955199373734528e-05,
+      "loss": 1.5654,
+      "step": 555
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.954210848382318e-05,
+      "loss": 1.5078,
+      "step": 560
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.953211585971524e-05,
+      "loss": 1.5417,
+      "step": 565
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.952201588667804e-05,
+      "loss": 1.5221,
+      "step": 570
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.951180858660089e-05,
+      "loss": 1.5447,
+      "step": 575
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.950149398160562e-05,
+      "loss": 1.5222,
+      "step": 580
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.949107209404665e-05,
+      "loss": 1.4868,
+      "step": 585
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.94805429465109e-05,
+      "loss": 1.5321,
+      "step": 590
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.946990656181781e-05,
+      "loss": 1.5181,
+      "step": 595
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.945916296301913e-05,
+      "loss": 1.5093,
+      "step": 600
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.944831217339906e-05,
+      "loss": 1.4971,
+      "step": 605
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.943735421647404e-05,
+      "loss": 1.5206,
+      "step": 610
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.94262891159928e-05,
+      "loss": 1.5135,
+      "step": 615
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.941511689593633e-05,
+      "loss": 1.4839,
+      "step": 620
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.940383758051769e-05,
+      "loss": 1.5178,
+      "step": 625
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.939245119418207e-05,
+      "loss": 1.5073,
+      "step": 630
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.938095776160675e-05,
+      "loss": 1.4663,
+      "step": 635
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.936935730770093e-05,
+      "loss": 1.507,
+      "step": 640
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.935764985760583e-05,
+      "loss": 1.4395,
+      "step": 645
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.934583543669453e-05,
+      "loss": 1.4925,
+      "step": 650
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.933391407057195e-05,
+      "loss": 1.4688,
+      "step": 655
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.932188578507476e-05,
+      "loss": 1.4939,
+      "step": 660
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.930975060627137e-05,
+      "loss": 1.4768,
+      "step": 665
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.929750856046187e-05,
+      "loss": 1.5037,
+      "step": 670
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.928515967417793e-05,
+      "loss": 1.5088,
+      "step": 675
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.92727039741828e-05,
+      "loss": 1.4859,
+      "step": 680
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.92601414874712e-05,
+      "loss": 1.4738,
+      "step": 685
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.924747224126932e-05,
+      "loss": 1.4612,
+      "step": 690
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.923469626303466e-05,
+      "loss": 1.4824,
+      "step": 695
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.922181358045607e-05,
+      "loss": 1.4828,
+      "step": 700
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.920882422145372e-05,
+      "loss": 1.4762,
+      "step": 705
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.919572821417886e-05,
+      "loss": 1.4874,
+      "step": 710
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.918252558701396e-05,
+      "loss": 1.4858,
+      "step": 715
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.916921636857253e-05,
+      "loss": 1.428,
+      "step": 720
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.915580058769909e-05,
+      "loss": 1.4686,
+      "step": 725
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.91422782734691e-05,
+      "loss": 1.4958,
+      "step": 730
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.912864945518893e-05,
+      "loss": 1.4865,
+      "step": 735
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.911491416239578e-05,
+      "loss": 1.4329,
+      "step": 740
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.910107242485756e-05,
+      "loss": 1.4989,
+      "step": 745
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.908712427257291e-05,
+      "loss": 1.4793,
+      "step": 750
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.90730697357711e-05,
+      "loss": 1.4498,
+      "step": 755
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.905890884491195e-05,
+      "loss": 1.4575,
+      "step": 760
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.904464163068578e-05,
+      "loss": 1.4614,
+      "step": 765
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.903026812401333e-05,
+      "loss": 1.4536,
+      "step": 770
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.901578835604571e-05,
+      "loss": 1.4457,
+      "step": 775
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.900120235816435e-05,
+      "loss": 1.4414,
+      "step": 780
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.898651016198086e-05,
+      "loss": 1.4442,
+      "step": 785
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.897171179933707e-05,
+      "loss": 1.4697,
+      "step": 790
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.895680730230483e-05,
+      "loss": 1.4686,
+      "step": 795
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.894179670318606e-05,
+      "loss": 1.4438,
+      "step": 800
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.892668003451264e-05,
+      "loss": 1.4462,
+      "step": 805
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.891145732904627e-05,
+      "loss": 1.445,
+      "step": 810
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.889612861977853e-05,
+      "loss": 1.4691,
+      "step": 815
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.88806939399307e-05,
+      "loss": 1.4118,
+      "step": 820
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.88651533229537e-05,
+      "loss": 1.4472,
+      "step": 825
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.884950680252811e-05,
+      "loss": 1.4528,
+      "step": 830
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.883375441256397e-05,
+      "loss": 1.4371,
+      "step": 835
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.881789618720081e-05,
+      "loss": 1.4241,
+      "step": 840
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.880193216080749e-05,
+      "loss": 1.4516,
+      "step": 845
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.878586236798222e-05,
+      "loss": 1.4063,
+      "step": 850
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.876968684355237e-05,
+      "loss": 1.4011,
+      "step": 855
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.875340562257453e-05,
+      "loss": 1.4601,
+      "step": 860
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.87370187403343e-05,
+      "loss": 1.412,
+      "step": 865
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.872052623234632e-05,
+      "loss": 1.4153,
+      "step": 870
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.87039281343541e-05,
+      "loss": 1.4465,
+      "step": 875
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.868722448233004e-05,
+      "loss": 1.4635,
+      "step": 880
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.867041531247525e-05,
+      "loss": 1.4641,
+      "step": 885
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.865350066121961e-05,
+      "loss": 1.4379,
+      "step": 890
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.863648056522152e-05,
+      "loss": 1.4262,
+      "step": 895
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.861935506136793e-05,
+      "loss": 1.458,
+      "step": 900
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.860212418677427e-05,
+      "loss": 1.4351,
+      "step": 905
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.85847879787843e-05,
+      "loss": 1.431,
+      "step": 910
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.856734647497006e-05,
+      "loss": 1.4258,
+      "step": 915
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.854979971313182e-05,
+      "loss": 1.4722,
+      "step": 920
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.853214773129796e-05,
+      "loss": 1.4278,
+      "step": 925
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.85143905677249e-05,
+      "loss": 1.4262,
+      "step": 930
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.849652826089701e-05,
+      "loss": 1.4168,
+      "step": 935
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.847856084952653e-05,
+      "loss": 1.3828,
+      "step": 940
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.846048837255353e-05,
+      "loss": 1.3864,
+      "step": 945
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.844231086914571e-05,
+      "loss": 1.4595,
+      "step": 950
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.842402837869843e-05,
+      "loss": 1.4573,
+      "step": 955
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.84056409408346e-05,
+      "loss": 1.4642,
+      "step": 960
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.838714859540458e-05,
+      "loss": 1.3841,
+      "step": 965
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.836855138248605e-05,
+      "loss": 1.4043,
+      "step": 970
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.834984934238398e-05,
+      "loss": 1.3911,
+      "step": 975
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.833104251563056e-05,
+      "loss": 1.4313,
+      "step": 980
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.831213094298506e-05,
+      "loss": 1.4181,
+      "step": 985
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.829311466543373e-05,
+      "loss": 1.3941,
+      "step": 990
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.827399372418978e-05,
+      "loss": 1.413,
+      "step": 995
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.825476816069326e-05,
+      "loss": 1.3816,
+      "step": 1000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.823543801661093e-05,
+      "loss": 1.4054,
+      "step": 1005
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.821600333383625e-05,
+      "loss": 1.4134,
+      "step": 1010
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.819646415448918e-05,
+      "loss": 1.4438,
+      "step": 1015
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.817682052091618e-05,
+      "loss": 1.4151,
+      "step": 1020
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.815707247569012e-05,
+      "loss": 1.4188,
+      "step": 1025
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.813722006161013e-05,
+      "loss": 1.4147,
+      "step": 1030
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.811726332170153e-05,
+      "loss": 1.4014,
+      "step": 1035
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.809720229921572e-05,
+      "loss": 1.3802,
+      "step": 1040
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.807703703763015e-05,
+      "loss": 1.3704,
+      "step": 1045
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.805676758064821e-05,
+      "loss": 1.4208,
+      "step": 1050
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.8036393972199e-05,
+      "loss": 1.4624,
+      "step": 1055
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.801591625643745e-05,
+      "loss": 1.3555,
+      "step": 1060
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.799533447774404e-05,
+      "loss": 1.41,
+      "step": 1065
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.797464868072488e-05,
+      "loss": 1.4434,
+      "step": 1070
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.795385891021139e-05,
+      "loss": 1.4145,
+      "step": 1075
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.79329652112604e-05,
+      "loss": 1.4278,
+      "step": 1080
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.7911967629154e-05,
+      "loss": 1.4084,
+      "step": 1085
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.789086620939936e-05,
+      "loss": 1.4019,
+      "step": 1090
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.786966099772874e-05,
+      "loss": 1.397,
+      "step": 1095
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.784835204009932e-05,
+      "loss": 1.3791,
+      "step": 1100
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.782693938269314e-05,
+      "loss": 1.4358,
+      "step": 1105
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.780542307191698e-05,
+      "loss": 1.3879,
+      "step": 1110
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.778380315440225e-05,
+      "loss": 1.3909,
+      "step": 1115
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.77620796770049e-05,
+      "loss": 1.4191,
+      "step": 1120
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.774025268680538e-05,
+      "loss": 1.3829,
+      "step": 1125
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.771832223110839e-05,
+      "loss": 1.4481,
+      "step": 1130
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.769628835744293e-05,
+      "loss": 1.382,
+      "step": 1135
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.76741511135621e-05,
+      "loss": 1.4017,
+      "step": 1140
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.765191054744305e-05,
+      "loss": 1.3549,
+      "step": 1145
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.762956670728685e-05,
+      "loss": 1.4065,
+      "step": 1150
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.76071196415184e-05,
+      "loss": 1.3903,
+      "step": 1155
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.758456939878629e-05,
+      "loss": 1.4029,
+      "step": 1160
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.756191602796276e-05,
+      "loss": 1.3629,
+      "step": 1165
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.753915957814352e-05,
+      "loss": 1.3874,
+      "step": 1170
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.75163000986477e-05,
+      "loss": 1.3745,
+      "step": 1175
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.74933376390177e-05,
+      "loss": 1.3811,
+      "step": 1180
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.747027224901913e-05,
+      "loss": 1.3938,
+      "step": 1185
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.744710397864067e-05,
+      "loss": 1.441,
+      "step": 1190
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.742383287809399e-05,
+      "loss": 1.385,
+      "step": 1195
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.740045899781352e-05,
+      "loss": 1.37,
+      "step": 1200
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.737698238845659e-05,
+      "loss": 1.4519,
+      "step": 1205
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.735340310090307e-05,
+      "loss": 1.3702,
+      "step": 1210
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.732972118625536e-05,
+      "loss": 1.3513,
+      "step": 1215
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.730593669583836e-05,
+      "loss": 1.3677,
+      "step": 1220
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.728204968119916e-05,
+      "loss": 1.3971,
+      "step": 1225
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.725806019410717e-05,
+      "loss": 1.4025,
+      "step": 1230
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.723396828655378e-05,
+      "loss": 1.347,
+      "step": 1235
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.720977401075242e-05,
+      "loss": 1.373,
+      "step": 1240
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.718547741913833e-05,
+      "loss": 1.3881,
+      "step": 1245
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.716107856436855e-05,
+      "loss": 1.408,
+      "step": 1250
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.713657749932172e-05,
+      "loss": 1.4115,
+      "step": 1255
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.711197427709796e-05,
+      "loss": 1.4012,
+      "step": 1260
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.708726895101886e-05,
+      "loss": 1.3753,
+      "step": 1265
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.706246157462726e-05,
+      "loss": 1.4148,
+      "step": 1270
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.703755220168715e-05,
+      "loss": 1.3844,
+      "step": 1275
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.701254088618362e-05,
+      "loss": 1.3881,
+      "step": 1280
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.698742768232267e-05,
+      "loss": 1.3378,
+      "step": 1285
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.696221264453109e-05,
+      "loss": 1.3523,
+      "step": 1290
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.693689582745644e-05,
+      "loss": 1.3995,
+      "step": 1295
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.69114772859668e-05,
+      "loss": 1.3719,
+      "step": 1300
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.688595707515076e-05,
+      "loss": 1.3505,
+      "step": 1305
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.686033525031719e-05,
+      "loss": 1.3584,
+      "step": 1310
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.683461186699525e-05,
+      "loss": 1.3668,
+      "step": 1315
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.680878698093417e-05,
+      "loss": 1.3936,
+      "step": 1320
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.678286064810316e-05,
+      "loss": 1.3823,
+      "step": 1325
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.675683292469132e-05,
+      "loss": 1.3751,
+      "step": 1330
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.673070386710745e-05,
+      "loss": 1.3899,
+      "step": 1335
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.670447353198e-05,
+      "loss": 1.385,
+      "step": 1340
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.667814197615692e-05,
+      "loss": 1.402,
+      "step": 1345
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.665170925670548e-05,
+      "loss": 1.3709,
+      "step": 1350
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.662517543091225e-05,
+      "loss": 1.348,
+      "step": 1355
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.659854055628291e-05,
+      "loss": 1.3954,
+      "step": 1360
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.657180469054213e-05,
+      "loss": 1.3816,
+      "step": 1365
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.654496789163345e-05,
+      "loss": 1.3648,
+      "step": 1370
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.651803021771917e-05,
+      "loss": 1.3939,
+      "step": 1375
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.649099172718021e-05,
+      "loss": 1.3361,
+      "step": 1380
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.6463852478616e-05,
+      "loss": 1.3986,
+      "step": 1385
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.643661253084431e-05,
+      "loss": 1.416,
+      "step": 1390
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.640927194290116e-05,
+      "loss": 1.3805,
+      "step": 1395
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.638183077404069e-05,
+      "loss": 1.3682,
+      "step": 1400
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.635428908373504e-05,
+      "loss": 1.3848,
+      "step": 1405
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.632664693167416e-05,
+      "loss": 1.3467,
+      "step": 1410
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.62989043777658e-05,
+      "loss": 1.4141,
+      "step": 1415
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.627106148213522e-05,
+      "loss": 1.3765,
+      "step": 1420
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.62431183051252e-05,
+      "loss": 1.3604,
+      "step": 1425
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.621507490729585e-05,
+      "loss": 1.3664,
+      "step": 1430
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.618693134942448e-05,
+      "loss": 1.353,
+      "step": 1435
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.615868769250546e-05,
+      "loss": 1.3409,
+      "step": 1440
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.613034399775016e-05,
+      "loss": 1.3811,
+      "step": 1445
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.610190032658663e-05,
+      "loss": 1.3831,
+      "step": 1450
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.607335674065976e-05,
+      "loss": 1.3528,
+      "step": 1455
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.604471330183083e-05,
+      "loss": 1.4102,
+      "step": 1460
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.601597007217762e-05,
+      "loss": 1.3509,
+      "step": 1465
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.598712711399416e-05,
+      "loss": 1.3679,
+      "step": 1470
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.595818448979061e-05,
+      "loss": 1.3578,
+      "step": 1475
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.592914226229314e-05,
+      "loss": 1.3599,
+      "step": 1480
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.590000049444377e-05,
+      "loss": 1.3833,
+      "step": 1485
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.587075924940028e-05,
+      "loss": 1.3838,
+      "step": 1490
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.584141859053602e-05,
+      "loss": 1.3662,
+      "step": 1495
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.581197858143978e-05,
+      "loss": 1.3331,
+      "step": 1500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.57824392859157e-05,
+      "loss": 1.3894,
+      "step": 1505
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.575280076798309e-05,
+      "loss": 1.3422,
+      "step": 1510
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.57230630918763e-05,
+      "loss": 1.3404,
+      "step": 1515
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.569322632204458e-05,
+      "loss": 1.3372,
+      "step": 1520
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.566329052315194e-05,
+      "loss": 1.3885,
+      "step": 1525
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.563325576007701e-05,
+      "loss": 1.3676,
+      "step": 1530
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.560312209791292e-05,
+      "loss": 1.329,
+      "step": 1535
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.557288960196707e-05,
+      "loss": 1.3785,
+      "step": 1540
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.554255833776119e-05,
+      "loss": 1.335,
+      "step": 1545
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.551212837103092e-05,
+      "loss": 1.3726,
+      "step": 1550
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.548159976772592e-05,
+      "loss": 1.3281,
+      "step": 1555
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.545097259400958e-05,
+      "loss": 1.3793,
+      "step": 1560
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.542024691625891e-05,
+      "loss": 1.384,
+      "step": 1565
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.538942280106443e-05,
+      "loss": 1.3819,
+      "step": 1570
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.535850031522996e-05,
+      "loss": 1.3203,
+      "step": 1575
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.53274795257726e-05,
+      "loss": 1.3764,
+      "step": 1580
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.529636049992234e-05,
+      "loss": 1.3648,
+      "step": 1585
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.526514330512225e-05,
+      "loss": 1.3029,
+      "step": 1590
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.523382800902804e-05,
+      "loss": 1.3373,
+      "step": 1595
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.520241467950811e-05,
+      "loss": 1.3265,
+      "step": 1600
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.517090338464325e-05,
+      "loss": 1.3161,
+      "step": 1605
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.513929419272662e-05,
+      "loss": 1.3528,
+      "step": 1610
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.510758717226351e-05,
+      "loss": 1.2851,
+      "step": 1615
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.507578239197126e-05,
+      "loss": 1.3991,
+      "step": 1620
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.504387992077907e-05,
+      "loss": 1.3667,
+      "step": 1625
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.501187982782785e-05,
+      "loss": 1.3757,
+      "step": 1630
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.497978218247012e-05,
+      "loss": 1.3963,
+      "step": 1635
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.494758705426978e-05,
+      "loss": 1.3473,
+      "step": 1640
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.4915294513002e-05,
+      "loss": 1.3945,
+      "step": 1645
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.48829046286531e-05,
+      "loss": 1.3694,
+      "step": 1650
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.485041747142034e-05,
+      "loss": 1.3352,
+      "step": 1655
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.481783311171183e-05,
+      "loss": 1.3585,
+      "step": 1660
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.47851516201463e-05,
+      "loss": 1.3317,
+      "step": 1665
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.475237306755302e-05,
+      "loss": 1.357,
+      "step": 1670
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.47194975249716e-05,
+      "loss": 1.3254,
+      "step": 1675
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.468652506365187e-05,
+      "loss": 1.3411,
+      "step": 1680
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.465345575505367e-05,
+      "loss": 1.3505,
+      "step": 1685
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.46202896708468e-05,
+      "loss": 1.3743,
+      "step": 1690
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.458702688291073e-05,
+      "loss": 1.3383,
+      "step": 1695
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.455366746333454e-05,
+      "loss": 1.3388,
+      "step": 1700
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.452021148441675e-05,
+      "loss": 1.3664,
+      "step": 1705
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.448665901866514e-05,
+      "loss": 1.3435,
+      "step": 1710
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.445301013879657e-05,
+      "loss": 1.3537,
+      "step": 1715
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.441926491773691e-05,
+      "loss": 1.3667,
+      "step": 1720
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.438542342862076e-05,
+      "loss": 1.3364,
+      "step": 1725
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.435148574479144e-05,
+      "loss": 1.319,
+      "step": 1730
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.43174519398007e-05,
+      "loss": 1.3481,
+      "step": 1735
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.428332208740857e-05,
+      "loss": 1.3649,
+      "step": 1740
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.424909626158332e-05,
+      "loss": 1.3461,
+      "step": 1745
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.421477453650118e-05,
+      "loss": 1.356,
+      "step": 1750
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.418035698654621e-05,
+      "loss": 1.3286,
+      "step": 1755
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.414584368631019e-05,
+      "loss": 1.314,
+      "step": 1760
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.411123471059233e-05,
+      "loss": 1.3308,
+      "step": 1765
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.407653013439928e-05,
+      "loss": 1.352,
+      "step": 1770
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.404173003294485e-05,
+      "loss": 1.3419,
+      "step": 1775
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.400683448164987e-05,
+      "loss": 1.3682,
+      "step": 1780
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.397184355614206e-05,
+      "loss": 1.3423,
+      "step": 1785
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.393675733225578e-05,
+      "loss": 1.3068,
+      "step": 1790
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.390157588603201e-05,
+      "loss": 1.3657,
+      "step": 1795
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.386629929371804e-05,
+      "loss": 1.3624,
+      "step": 1800
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.38309276317674e-05,
+      "loss": 1.3353,
+      "step": 1805
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.379546097683962e-05,
+      "loss": 1.3604,
+      "step": 1810
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.375989940580014e-05,
+      "loss": 1.3696,
+      "step": 1815
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.372424299572013e-05,
+      "loss": 1.3718,
+      "step": 1820
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.368849182387625e-05,
+      "loss": 1.3601,
+      "step": 1825
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.365264596775051e-05,
+      "loss": 1.3819,
+      "step": 1830
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.361670550503024e-05,
+      "loss": 1.333,
+      "step": 1835
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.35806705136077e-05,
+      "loss": 1.3172,
+      "step": 1840
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.354454107158003e-05,
+      "loss": 1.34,
+      "step": 1845
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.350831725724916e-05,
+      "loss": 1.3823,
+      "step": 1850
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.34719991491214e-05,
+      "loss": 1.3403,
+      "step": 1855
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.343558682590756e-05,
+      "loss": 1.3131,
+      "step": 1860
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.339908036652255e-05,
+      "loss": 1.2946,
+      "step": 1865
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.336247985008534e-05,
+      "loss": 1.3053,
+      "step": 1870
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.332578535591872e-05,
+      "loss": 1.313,
+      "step": 1875
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.328899696354918e-05,
+      "loss": 1.3495,
+      "step": 1880
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.32521147527067e-05,
+      "loss": 1.3487,
+      "step": 1885
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.321513880332458e-05,
+      "loss": 1.3428,
+      "step": 1890
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.317806919553927e-05,
+      "loss": 1.322,
+      "step": 1895
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.314090600969024e-05,
+      "loss": 1.3269,
+      "step": 1900
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.310364932631972e-05,
+      "loss": 1.3305,
+      "step": 1905
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.306629922617261e-05,
+      "loss": 1.3108,
+      "step": 1910
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.302885579019627e-05,
+      "loss": 1.3546,
+      "step": 1915
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.29913190995403e-05,
+      "loss": 1.3349,
+      "step": 1920
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.295368923555643e-05,
+      "loss": 1.3494,
+      "step": 1925
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.291596627979836e-05,
+      "loss": 1.3295,
+      "step": 1930
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.287815031402148e-05,
+      "loss": 1.3302,
+      "step": 1935
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.284024142018281e-05,
+      "loss": 1.311,
+      "step": 1940
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.280223968044073e-05,
+      "loss": 1.3141,
+      "step": 1945
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.276414517715484e-05,
+      "loss": 1.331,
+      "step": 1950
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.272595799288584e-05,
+      "loss": 1.373,
+      "step": 1955
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.268767821039521e-05,
+      "loss": 1.3349,
+      "step": 1960
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.264930591264518e-05,
+      "loss": 1.3243,
+      "step": 1965
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.261084118279847e-05,
+      "loss": 1.3466,
+      "step": 1970
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.25722841042181e-05,
+      "loss": 1.3006,
+      "step": 1975
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.253363476046725e-05,
+      "loss": 1.3251,
+      "step": 1980
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.249489323530908e-05,
+      "loss": 1.3236,
+      "step": 1985
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.245605961270649e-05,
+      "loss": 1.2814,
+      "step": 1990
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.241713397682202e-05,
+      "loss": 1.3518,
+      "step": 1995
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.23781164120176e-05,
+      "loss": 1.3397,
+      "step": 2000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.23390070028544e-05,
+      "loss": 1.3775,
+      "step": 2005
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.229980583409266e-05,
+      "loss": 1.3692,
+      "step": 2010
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.226051299069144e-05,
+      "loss": 1.3308,
+      "step": 2015
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.222112855780856e-05,
+      "loss": 1.3197,
+      "step": 2020
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.218165262080023e-05,
+      "loss": 1.3395,
+      "step": 2025
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.214208526522109e-05,
+      "loss": 1.322,
+      "step": 2030
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.21024265768238e-05,
+      "loss": 1.2909,
+      "step": 2035
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.206267664155907e-05,
+      "loss": 1.3124,
+      "step": 2040
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.202283554557525e-05,
+      "loss": 1.287,
+      "step": 2045
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.198290337521838e-05,
+      "loss": 1.3099,
+      "step": 2050
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.194288021703177e-05,
+      "loss": 1.3224,
+      "step": 2055
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.190276615775599e-05,
+      "loss": 1.3438,
+      "step": 2060
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.186256128432862e-05,
+      "loss": 1.2923,
+      "step": 2065
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.182226568388401e-05,
+      "loss": 1.3185,
+      "step": 2070
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.178187944375317e-05,
+      "loss": 1.3498,
+      "step": 2075
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.174140265146356e-05,
+      "loss": 1.3022,
+      "step": 2080
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.170083539473888e-05,
+      "loss": 1.314,
+      "step": 2085
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.166017776149887e-05,
+      "loss": 1.3123,
+      "step": 2090
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.161942983985916e-05,
+      "loss": 1.2835,
+      "step": 2095
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.157859171813107e-05,
+      "loss": 1.3162,
+      "step": 2100
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.153766348482138e-05,
+      "loss": 1.3363,
+      "step": 2105
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.149664522863217e-05,
+      "loss": 1.2964,
+      "step": 2110
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.145553703846064e-05,
+      "loss": 1.3859,
+      "step": 2115
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.141433900339887e-05,
+      "loss": 1.3518,
+      "step": 2120
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.13730512127337e-05,
+      "loss": 1.2937,
+      "step": 2125
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.133167375594647e-05,
+      "loss": 1.3514,
+      "step": 2130
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.129020672271283e-05,
+      "loss": 1.3189,
+      "step": 2135
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.12486502029026e-05,
+      "loss": 1.322,
+      "step": 2140
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.120700428657954e-05,
+      "loss": 1.2974,
+      "step": 2145
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.11652690640011e-05,
+      "loss": 1.3402,
+      "step": 2150
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.112344462561836e-05,
+      "loss": 1.2422,
+      "step": 2155
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.10815310620757e-05,
+      "loss": 1.3153,
+      "step": 2160
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.103952846421067e-05,
+      "loss": 1.3364,
+      "step": 2165
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.099743692305379e-05,
+      "loss": 1.3592,
+      "step": 2170
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.095525652982833e-05,
+      "loss": 1.3361,
+      "step": 2175
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.091298737595014e-05,
+      "loss": 1.3587,
+      "step": 2180
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.087062955302741e-05,
+      "loss": 1.3799,
+      "step": 2185
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.082818315286055e-05,
+      "loss": 1.3667,
+      "step": 2190
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.078564826744188e-05,
+      "loss": 1.3312,
+      "step": 2195
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.074302498895552e-05,
+      "loss": 1.3163,
+      "step": 2200
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.070031340977716e-05,
+      "loss": 1.2888,
+      "step": 2205
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.065751362247388e-05,
+      "loss": 1.3175,
+      "step": 2210
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.061462571980388e-05,
+      "loss": 1.3161,
+      "step": 2215
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.057164979471635e-05,
+      "loss": 1.3163,
+      "step": 2220
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.05285859403513e-05,
+      "loss": 1.3215,
+      "step": 2225
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.048543425003923e-05,
+      "loss": 1.3096,
+      "step": 2230
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.044219481730106e-05,
+      "loss": 1.2927,
+      "step": 2235
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.039886773584779e-05,
+      "loss": 1.3431,
+      "step": 2240
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.035545309958046e-05,
+      "loss": 1.3045,
+      "step": 2245
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.031195100258987e-05,
+      "loss": 1.3031,
+      "step": 2250
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.02683615391563e-05,
+      "loss": 1.3192,
+      "step": 2255
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.02246848037494e-05,
+      "loss": 1.3302,
+      "step": 2260
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.0180920891028e-05,
+      "loss": 1.3087,
+      "step": 2265
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.013706989583983e-05,
+      "loss": 1.291,
+      "step": 2270
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.009313191322137e-05,
+      "loss": 1.3093,
+      "step": 2275
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.00491070383976e-05,
+      "loss": 1.3234,
+      "step": 2280
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.000499536678182e-05,
+      "loss": 1.3329,
+      "step": 2285
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.996079699397547e-05,
+      "loss": 1.309,
+      "step": 2290
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.991651201576788e-05,
+      "loss": 1.3058,
+      "step": 2295
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.987214052813604e-05,
+      "loss": 1.3076,
+      "step": 2300
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.98276826272445e-05,
+      "loss": 1.2824,
+      "step": 2305
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.978313840944503e-05,
+      "loss": 1.2982,
+      "step": 2310
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.973850797127649e-05,
+      "loss": 1.3252,
+      "step": 2315
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.969379140946464e-05,
+      "loss": 1.307,
+      "step": 2320
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.964898882092178e-05,
+      "loss": 1.319,
+      "step": 2325
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.960410030274681e-05,
+      "loss": 1.3207,
+      "step": 2330
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.955912595222473e-05,
+      "loss": 1.319,
+      "step": 2335
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.951406586682662e-05,
+      "loss": 1.3162,
+      "step": 2340
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.946892014420935e-05,
+      "loss": 1.2752,
+      "step": 2345
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.942368888221545e-05,
+      "loss": 1.3014,
+      "step": 2350
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.937837217887273e-05,
+      "loss": 1.3255,
+      "step": 2355
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.933297013239424e-05,
+      "loss": 1.3247,
+      "step": 2360
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.928748284117798e-05,
+      "loss": 1.2847,
+      "step": 2365
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.924191040380671e-05,
+      "loss": 1.3403,
+      "step": 2370
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.91962529190477e-05,
+      "loss": 1.2964,
+      "step": 2375
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.915051048585256e-05,
+      "loss": 1.3115,
+      "step": 2380
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.910468320335697e-05,
+      "loss": 1.2945,
+      "step": 2385
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.905877117088054e-05,
+      "loss": 1.3632,
+      "step": 2390
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.901277448792656e-05,
+      "loss": 1.2907,
+      "step": 2395
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.896669325418172e-05,
+      "loss": 1.3439,
+      "step": 2400
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.892052756951604e-05,
+      "loss": 1.3015,
+      "step": 2405
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.887427753398248e-05,
+      "loss": 1.2839,
+      "step": 2410
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.882794324781686e-05,
+      "loss": 1.3056,
+      "step": 2415
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.87815248114376e-05,
+      "loss": 1.3167,
+      "step": 2420
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.873502232544546e-05,
+      "loss": 1.2938,
+      "step": 2425
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.868843589062339e-05,
+      "loss": 1.3241,
+      "step": 2430
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.864176560793625e-05,
+      "loss": 1.3518,
+      "step": 2435
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.859501157853066e-05,
+      "loss": 1.3045,
+      "step": 2440
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.85481739037347e-05,
+      "loss": 1.3045,
+      "step": 2445
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.850125268505774e-05,
+      "loss": 1.3036,
+      "step": 2450
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.845424802419025e-05,
+      "loss": 1.3013,
+      "step": 2455
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.840716002300347e-05,
+      "loss": 1.3226,
+      "step": 2460
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.835998878354931e-05,
+      "loss": 1.3381,
+      "step": 2465
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.831273440806009e-05,
+      "loss": 1.3097,
+      "step": 2470
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.826539699894825e-05,
+      "loss": 1.2881,
+      "step": 2475
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.821797665880625e-05,
+      "loss": 1.3076,
+      "step": 2480
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.817047349040626e-05,
+      "loss": 1.313,
+      "step": 2485
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.812288759669994e-05,
+      "loss": 1.292,
+      "step": 2490
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.807521908081822e-05,
+      "loss": 1.2571,
+      "step": 2495
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.802746804607118e-05,
+      "loss": 1.3165,
+      "step": 2500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.797963459594767e-05,
+      "loss": 1.3353,
+      "step": 2505
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.793171883411515e-05,
+      "loss": 1.296,
+      "step": 2510
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.788372086441949e-05,
+      "loss": 1.3302,
+      "step": 2515
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.783564079088477e-05,
+      "loss": 1.3092,
+      "step": 2520
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.778747871771292e-05,
+      "loss": 1.284,
+      "step": 2525
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.773923474928365e-05,
+      "loss": 1.2885,
+      "step": 2530
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.769090899015414e-05,
+      "loss": 1.313,
+      "step": 2535
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.764250154505885e-05,
+      "loss": 1.3561,
+      "step": 2540
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.759401251890921e-05,
+      "loss": 1.3018,
+      "step": 2545
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.754544201679353e-05,
+      "loss": 1.3002,
+      "step": 2550
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.749679014397669e-05,
+      "loss": 1.309,
+      "step": 2555
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.744805700589989e-05,
+      "loss": 1.298,
+      "step": 2560
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.739924270818045e-05,
+      "loss": 1.2911,
+      "step": 2565
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.735034735661162e-05,
+      "loss": 1.3262,
+      "step": 2570
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.73013710571623e-05,
+      "loss": 1.3155,
+      "step": 2575
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.725231391597681e-05,
+      "loss": 1.294,
+      "step": 2580
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.72031760393747e-05,
+      "loss": 1.3269,
+      "step": 2585
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.715395753385048e-05,
+      "loss": 1.2686,
+      "step": 2590
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.710465850607337e-05,
+      "loss": 1.3004,
+      "step": 2595
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.705527906288718e-05,
+      "loss": 1.3043,
+      "step": 2600
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.700581931130992e-05,
+      "loss": 1.2578,
+      "step": 2605
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.695627935853373e-05,
+      "loss": 1.2821,
+      "step": 2610
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.690665931192447e-05,
+      "loss": 1.3363,
+      "step": 2615
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.68569592790217e-05,
+      "loss": 1.2871,
+      "step": 2620
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.680717936753822e-05,
+      "loss": 1.3259,
+      "step": 2625
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.675731968536002e-05,
+      "loss": 1.2866,
+      "step": 2630
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.670738034054595e-05,
+      "loss": 1.3245,
+      "step": 2635
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.66573614413275e-05,
+      "loss": 1.346,
+      "step": 2640
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.66072630961086e-05,
+      "loss": 1.2979,
+      "step": 2645
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.655708541346533e-05,
+      "loss": 1.2942,
+      "step": 2650
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.650682850214572e-05,
+      "loss": 1.2954,
+      "step": 2655
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.645649247106955e-05,
+      "loss": 1.3032,
+      "step": 2660
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.640607742932802e-05,
+      "loss": 1.3004,
+      "step": 2665
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.635558348618359e-05,
+      "loss": 1.2858,
+      "step": 2670
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.630501075106972e-05,
+      "loss": 1.2779,
+      "step": 2675
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.625435933359062e-05,
+      "loss": 1.3242,
+      "step": 2680
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.620362934352109e-05,
+      "loss": 1.3095,
+      "step": 2685
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.615282089080609e-05,
+      "loss": 1.2871,
+      "step": 2690
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.610193408556073e-05,
+      "loss": 1.3356,
+      "step": 2695
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.605096903806991e-05,
+      "loss": 1.3052,
+      "step": 2700
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.599992585878812e-05,
+      "loss": 1.2827,
+      "step": 2705
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.594880465833908e-05,
+      "loss": 1.3148,
+      "step": 2710
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.589760554751573e-05,
+      "loss": 1.3064,
+      "step": 2715
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.584632863727982e-05,
+      "loss": 1.2904,
+      "step": 2720
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.579497403876166e-05,
+      "loss": 1.2962,
+      "step": 2725
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.574354186326001e-05,
+      "loss": 1.3187,
+      "step": 2730
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.56920322222417e-05,
+      "loss": 1.294,
+      "step": 2735
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.564044522734147e-05,
+      "loss": 1.3104,
+      "step": 2740
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.558878099036169e-05,
+      "loss": 1.2927,
+      "step": 2745
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.55370396232722e-05,
+      "loss": 1.2776,
+      "step": 2750
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.54852212382099e-05,
+      "loss": 1.2716,
+      "step": 2755
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.543332594747865e-05,
+      "loss": 1.2992,
+      "step": 2760
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.538135386354903e-05,
+      "loss": 1.3302,
+      "step": 2765
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.532930509905799e-05,
+      "loss": 1.265,
+      "step": 2770
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.527717976680869e-05,
+      "loss": 1.264,
+      "step": 2775
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.522497797977024e-05,
+      "loss": 1.2952,
+      "step": 2780
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.517269985107743e-05,
+      "loss": 1.3153,
+      "step": 2785
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.512034549403053e-05,
+      "loss": 1.3182,
+      "step": 2790
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.506791502209496e-05,
+      "loss": 1.3021,
+      "step": 2795
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.501540854890118e-05,
+      "loss": 1.3431,
+      "step": 2800
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.496282618824427e-05,
+      "loss": 1.3081,
+      "step": 2805
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.491016805408387e-05,
+      "loss": 1.2852,
+      "step": 2810
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.485743426054382e-05,
+      "loss": 1.3174,
+      "step": 2815
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.480462492191186e-05,
+      "loss": 1.3131,
+      "step": 2820
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.475174015263954e-05,
+      "loss": 1.2926,
+      "step": 2825
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.469878006734185e-05,
+      "loss": 1.3412,
+      "step": 2830
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.464574478079703e-05,
+      "loss": 1.286,
+      "step": 2835
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.459263440794627e-05,
+      "loss": 1.2941,
+      "step": 2840
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.453944906389353e-05,
+      "loss": 1.3166,
+      "step": 2845
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.448618886390522e-05,
+      "loss": 1.2807,
+      "step": 2850
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.443285392340998e-05,
+      "loss": 1.3257,
+      "step": 2855
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.437944435799848e-05,
+      "loss": 1.2925,
+      "step": 2860
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.432596028342303e-05,
+      "loss": 1.2969,
+      "step": 2865
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.427240181559754e-05,
+      "loss": 1.2709,
+      "step": 2870
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.421876907059707e-05,
+      "loss": 1.2694,
+      "step": 2875
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.416506216465765e-05,
+      "loss": 1.3055,
+      "step": 2880
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.411128121417609e-05,
+      "loss": 1.3141,
+      "step": 2885
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.405742633570961e-05,
+      "loss": 1.3149,
+      "step": 2890
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.400349764597571e-05,
+      "loss": 1.3001,
+      "step": 2895
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.394949526185185e-05,
+      "loss": 1.3147,
+      "step": 2900
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.389541930037516e-05,
+      "loss": 1.2889,
+      "step": 2905
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.384126987874228e-05,
+      "loss": 1.2822,
+      "step": 2910
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.378704711430901e-05,
+      "loss": 1.2842,
+      "step": 2915
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.373275112459016e-05,
+      "loss": 1.2641,
+      "step": 2920
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.36783820272592e-05,
+      "loss": 1.3062,
+      "step": 2925
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.362393994014805e-05,
+      "loss": 1.2688,
+      "step": 2930
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.356942498124681e-05,
+      "loss": 1.3275,
+      "step": 2935
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.35148372687035e-05,
+      "loss": 1.2904,
+      "step": 2940
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.34601769208239e-05,
+      "loss": 1.2769,
+      "step": 2945
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.340544405607111e-05,
+      "loss": 1.2798,
+      "step": 2950
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.335063879306543e-05,
+      "loss": 1.2952,
+      "step": 2955
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.329576125058406e-05,
+      "loss": 1.3038,
+      "step": 2960
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.324081154756087e-05,
+      "loss": 1.2482,
+      "step": 2965
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.318578980308609e-05,
+      "loss": 1.3026,
+      "step": 2970
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.31306961364061e-05,
+      "loss": 1.2946,
+      "step": 2975
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.307553066692314e-05,
+      "loss": 1.3095,
+      "step": 2980
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.302029351419511e-05,
+      "loss": 1.2737,
+      "step": 2985
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.29649847979352e-05,
+      "loss": 1.3048,
+      "step": 2990
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.290960463801174e-05,
+      "loss": 1.2649,
+      "step": 2995
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.28541531544479e-05,
+      "loss": 1.2815,
+      "step": 3000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.279863046742138e-05,
+      "loss": 1.3132,
+      "step": 3005
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.274303669726426e-05,
+      "loss": 1.3023,
+      "step": 3010
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.268737196446264e-05,
+      "loss": 1.2832,
+      "step": 3015
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.263163638965639e-05,
+      "loss": 1.3043,
+      "step": 3020
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.257583009363898e-05,
+      "loss": 1.2507,
+      "step": 3025
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.25199531973571e-05,
+      "loss": 1.2905,
+      "step": 3030
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.246400582191045e-05,
+      "loss": 1.2642,
+      "step": 3035
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.24079880885515e-05,
+      "loss": 1.2714,
+      "step": 3040
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.23519001186852e-05,
+      "loss": 1.3041,
+      "step": 3045
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.22957420338687e-05,
+      "loss": 1.3094,
+      "step": 3050
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.223951395581113e-05,
+      "loss": 1.2973,
+      "step": 3055
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.218321600637329e-05,
+      "loss": 1.2937,
+      "step": 3060
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.212684830756745e-05,
+      "loss": 1.2871,
+      "step": 3065
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.2070410981557e-05,
+      "loss": 1.3141,
+      "step": 3070
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.201390415065626e-05,
+      "loss": 1.2307,
+      "step": 3075
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.195732793733014e-05,
+      "loss": 1.3145,
+      "step": 3080
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.1900682464194e-05,
+      "loss": 1.2856,
+      "step": 3085
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.184396785401322e-05,
+      "loss": 1.2657,
+      "step": 3090
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.17871842297031e-05,
+      "loss": 1.272,
+      "step": 3095
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.173033171432841e-05,
+      "loss": 1.3025,
+      "step": 3100
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.167341043110333e-05,
+      "loss": 1.2909,
+      "step": 3105
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.1616420503391e-05,
+      "loss": 1.2624,
+      "step": 3110
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.155936205470338e-05,
+      "loss": 1.2986,
+      "step": 3115
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.15022352087009e-05,
+      "loss": 1.276,
+      "step": 3120
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.144504008919222e-05,
+      "loss": 1.2926,
+      "step": 3125
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.138777682013403e-05,
+      "loss": 1.2342,
+      "step": 3130
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.133044552563065e-05,
+      "loss": 1.2859,
+      "step": 3135
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.127304632993382e-05,
+      "loss": 1.3008,
+      "step": 3140
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.12155793574425e-05,
+      "loss": 1.2553,
+      "step": 3145
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.115804473270253e-05,
+      "loss": 1.3143,
+      "step": 3150
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.11004425804063e-05,
+      "loss": 1.2705,
+      "step": 3155
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.104277302539264e-05,
+      "loss": 1.2916,
+      "step": 3160
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.098503619264638e-05,
+      "loss": 1.2742,
+      "step": 3165
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.092723220729825e-05,
+      "loss": 1.2757,
+      "step": 3170
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.086936119462441e-05,
+      "loss": 1.3152,
+      "step": 3175
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.081142328004637e-05,
+      "loss": 1.2662,
+      "step": 3180
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.075341858913059e-05,
+      "loss": 1.2715,
+      "step": 3185
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.069534724758827e-05,
+      "loss": 1.2626,
+      "step": 3190
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.063720938127506e-05,
+      "loss": 1.2605,
+      "step": 3195
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.057900511619076e-05,
+      "loss": 1.2866,
+      "step": 3200
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.052073457847913e-05,
+      "loss": 1.2929,
+      "step": 3205
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.046239789442749e-05,
+      "loss": 1.2953,
+      "step": 3210
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.040399519046656e-05,
+      "loss": 1.295,
+      "step": 3215
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.034552659317012e-05,
+      "loss": 1.2681,
+      "step": 3220
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.028699222925478e-05,
+      "loss": 1.3386,
+      "step": 3225
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.02283922255797e-05,
+      "loss": 1.2676,
+      "step": 3230
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.016972670914624e-05,
+      "loss": 1.3058,
+      "step": 3235
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.011099580709778e-05,
+      "loss": 1.3105,
+      "step": 3240
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.00521996467194e-05,
+      "loss": 1.2635,
+      "step": 3245
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.999333835543763e-05,
+      "loss": 1.29,
+      "step": 3250
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.993441206082011e-05,
+      "loss": 1.3163,
+      "step": 3255
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.987542089057542e-05,
+      "loss": 1.2676,
+      "step": 3260
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.98163649725527e-05,
+      "loss": 1.2801,
+      "step": 3265
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.975724443474143e-05,
+      "loss": 1.2913,
+      "step": 3270
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.969805940527112e-05,
+      "loss": 1.2806,
+      "step": 3275
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.963881001241107e-05,
+      "loss": 1.3015,
+      "step": 3280
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.957949638457009e-05,
+      "loss": 1.2442,
+      "step": 3285
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.952011865029614e-05,
+      "loss": 1.2956,
+      "step": 3290
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.946067693827617e-05,
+      "loss": 1.3245,
+      "step": 3295
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.940117137733579e-05,
+      "loss": 1.2487,
+      "step": 3300
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.934160209643896e-05,
+      "loss": 1.2737,
+      "step": 3305
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.928196922468772e-05,
+      "loss": 1.3004,
+      "step": 3310
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.9222272891322e-05,
+      "loss": 1.2892,
+      "step": 3315
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.916251322571918e-05,
+      "loss": 1.3061,
+      "step": 3320
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.910269035739397e-05,
+      "loss": 1.2979,
+      "step": 3325
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.904280441599801e-05,
+      "loss": 1.2976,
+      "step": 3330
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.898285553131967e-05,
+      "loss": 1.2707,
+      "step": 3335
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.892284383328367e-05,
+      "loss": 1.3109,
+      "step": 3340
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.886276945195099e-05,
+      "loss": 1.3075,
+      "step": 3345
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.88026325175183e-05,
+      "loss": 1.2973,
+      "step": 3350
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.874243316031796e-05,
+      "loss": 1.2827,
+      "step": 3355
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.868217151081755e-05,
+      "loss": 1.2985,
+      "step": 3360
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.86218476996197e-05,
+      "loss": 1.2577,
+      "step": 3365
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.856146185746175e-05,
+      "loss": 1.3309,
+      "step": 3370
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.850101411521544e-05,
+      "loss": 1.2704,
+      "step": 3375
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.844050460388671e-05,
+      "loss": 1.3145,
+      "step": 3380
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.837993345461534e-05,
+      "loss": 1.2716,
+      "step": 3385
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.831930079867469e-05,
+      "loss": 1.2927,
+      "step": 3390
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.825860676747145e-05,
+      "loss": 1.2723,
+      "step": 3395
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.819785149254532e-05,
+      "loss": 1.2963,
+      "step": 3400
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.813703510556873e-05,
+      "loss": 1.2836,
+      "step": 3405
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.807615773834652e-05,
+      "loss": 1.278,
+      "step": 3410
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.801521952281574e-05,
+      "loss": 1.2725,
+      "step": 3415
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.795422059104527e-05,
+      "loss": 1.2562,
+      "step": 3420
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.789316107523564e-05,
+      "loss": 1.3309,
+      "step": 3425
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.78320411077186e-05,
+      "loss": 1.2532,
+      "step": 3430
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.777086082095697e-05,
+      "loss": 1.2651,
+      "step": 3435
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.77096203475443e-05,
+      "loss": 1.2683,
+      "step": 3440
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.764831982020455e-05,
+      "loss": 1.2543,
+      "step": 3445
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.758695937179185e-05,
+      "loss": 1.2705,
+      "step": 3450
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.752553913529018e-05,
+      "loss": 1.2843,
+      "step": 3455
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.746405924381313e-05,
+      "loss": 1.272,
+      "step": 3460
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.740251983060352e-05,
+      "loss": 1.3018,
+      "step": 3465
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.734092102903323e-05,
+      "loss": 1.2719,
+      "step": 3470
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.727926297260282e-05,
+      "loss": 1.2521,
+      "step": 3475
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.721754579494127e-05,
+      "loss": 1.2871,
+      "step": 3480
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.71557696298057e-05,
+      "loss": 1.299,
+      "step": 3485
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.709393461108107e-05,
+      "loss": 1.2912,
+      "step": 3490
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.703204087277988e-05,
+      "loss": 1.2715,
+      "step": 3495
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.697008854904191e-05,
+      "loss": 1.3105,
+      "step": 3500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.690807777413387e-05,
+      "loss": 1.2319,
+      "step": 3505
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.68460086824492e-05,
+      "loss": 1.2762,
+      "step": 3510
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.678388140850769e-05,
+      "loss": 1.3042,
+      "step": 3515
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.672169608695525e-05,
+      "loss": 1.26,
+      "step": 3520
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.66594528525636e-05,
+      "loss": 1.2597,
+      "step": 3525
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.659715184022994e-05,
+      "loss": 1.2822,
+      "step": 3530
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.65347931849767e-05,
+      "loss": 1.2773,
+      "step": 3535
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.647237702195123e-05,
+      "loss": 1.2959,
+      "step": 3540
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.640990348642556e-05,
+      "loss": 1.2554,
+      "step": 3545
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.634737271379603e-05,
+      "loss": 1.3193,
+      "step": 3550
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.628478483958299e-05,
+      "loss": 1.2536,
+      "step": 3555
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.622213999943062e-05,
+      "loss": 1.2768,
+      "step": 3560
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.61594383291065e-05,
+      "loss": 1.2716,
+      "step": 3565
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.609667996450141e-05,
+      "loss": 1.2797,
+      "step": 3570
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.603386504162897e-05,
+      "loss": 1.3076,
+      "step": 3575
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.59709936966254e-05,
+      "loss": 1.2725,
+      "step": 3580
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.59080660657492e-05,
+      "loss": 1.2427,
+      "step": 3585
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.584508228538085e-05,
+      "loss": 1.2771,
+      "step": 3590
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.57820424920225e-05,
+      "loss": 1.2419,
+      "step": 3595
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.571894682229775e-05,
+      "loss": 1.2572,
+      "step": 3600
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.565579541295125e-05,
+      "loss": 1.2778,
+      "step": 3605
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.559258840084848e-05,
+      "loss": 1.2599,
+      "step": 3610
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.552932592297536e-05,
+      "loss": 1.3102,
+      "step": 3615
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.546600811643816e-05,
+      "loss": 1.2691,
+      "step": 3620
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.54026351184629e-05,
+      "loss": 1.232,
+      "step": 3625
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.533920706639531e-05,
+      "loss": 1.2356,
+      "step": 3630
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.527572409770043e-05,
+      "loss": 1.3063,
+      "step": 3635
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.521218634996226e-05,
+      "loss": 1.2793,
+      "step": 3640
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.51485939608836e-05,
+      "loss": 1.2891,
+      "step": 3645
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.508494706828564e-05,
+      "loss": 1.2753,
+      "step": 3650
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.502124581010766e-05,
+      "loss": 1.2505,
+      "step": 3655
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.49574903244068e-05,
+      "loss": 1.275,
+      "step": 3660
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.489368074935773e-05,
+      "loss": 1.253,
+      "step": 3665
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.482981722325232e-05,
+      "loss": 1.2401,
+      "step": 3670
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.476589988449939e-05,
+      "loss": 1.2706,
+      "step": 3675
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.470192887162435e-05,
+      "loss": 1.2634,
+      "step": 3680
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.463790432326898e-05,
+      "loss": 1.2786,
+      "step": 3685
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.457382637819108e-05,
+      "loss": 1.2728,
+      "step": 3690
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.450969517526412e-05,
+      "loss": 1.2606,
+      "step": 3695
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.444551085347707e-05,
+      "loss": 1.2861,
+      "step": 3700
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.438127355193396e-05,
+      "loss": 1.3247,
+      "step": 3705
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.43169834098537e-05,
+      "loss": 1.2594,
+      "step": 3710
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.425264056656966e-05,
+      "loss": 1.2662,
+      "step": 3715
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.418824516152943e-05,
+      "loss": 1.2364,
+      "step": 3720
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.41237973342946e-05,
+      "loss": 1.2748,
+      "step": 3725
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.405929722454026e-05,
+      "loss": 1.2504,
+      "step": 3730
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.399474497205488e-05,
+      "loss": 1.2507,
+      "step": 3735
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.393014071673992e-05,
+      "loss": 1.2794,
+      "step": 3740
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.386548459860953e-05,
+      "loss": 1.2563,
+      "step": 3745
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.380077675779027e-05,
+      "loss": 1.246,
+      "step": 3750
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.373601733452081e-05,
+      "loss": 1.27,
+      "step": 3755
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.36712064691516e-05,
+      "loss": 1.2502,
+      "step": 3760
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.360634430214456e-05,
+      "loss": 1.3012,
+      "step": 3765
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.354143097407283e-05,
+      "loss": 1.303,
+      "step": 3770
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.34764666256204e-05,
+      "loss": 1.2909,
+      "step": 3775
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.341145139758185e-05,
+      "loss": 1.2424,
+      "step": 3780
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.334638543086203e-05,
+      "loss": 1.2309,
+      "step": 3785
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.328126886647575e-05,
+      "loss": 1.2646,
+      "step": 3790
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.321610184554747e-05,
+      "loss": 1.2727,
+      "step": 3795
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.315088450931103e-05,
+      "loss": 1.2433,
+      "step": 3800
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.308561699910923e-05,
+      "loss": 1.2953,
+      "step": 3805
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.302029945639377e-05,
+      "loss": 1.2205,
+      "step": 3810
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.295493202272462e-05,
+      "loss": 1.2718,
+      "step": 3815
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.288951483976998e-05,
+      "loss": 1.2338,
+      "step": 3820
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.282404804930583e-05,
+      "loss": 1.259,
+      "step": 3825
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.275853179321565e-05,
+      "loss": 1.275,
+      "step": 3830
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.269296621349016e-05,
+      "loss": 1.2848,
+      "step": 3835
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.262735145222696e-05,
+      "loss": 1.2792,
+      "step": 3840
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.25616876516302e-05,
+      "loss": 1.2825,
+      "step": 3845
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.249597495401043e-05,
+      "loss": 1.2969,
+      "step": 3850
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.243021350178402e-05,
+      "loss": 1.2341,
+      "step": 3855
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.236440343747313e-05,
+      "loss": 1.2681,
+      "step": 3860
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.229854490370518e-05,
+      "loss": 1.2729,
+      "step": 3865
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.223263804321269e-05,
+      "loss": 1.3192,
+      "step": 3870
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.21666829988329e-05,
+      "loss": 1.2644,
+      "step": 3875
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.21006799135075e-05,
+      "loss": 1.276,
+      "step": 3880
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.203462893028224e-05,
+      "loss": 1.2637,
+      "step": 3885
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.196853019230676e-05,
+      "loss": 1.2904,
+      "step": 3890
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.190238384283412e-05,
+      "loss": 1.2893,
+      "step": 3895
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.183619002522062e-05,
+      "loss": 1.284,
+      "step": 3900
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.176994888292538e-05,
+      "loss": 1.2313,
+      "step": 3905
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.170366055951017e-05,
+      "loss": 1.2534,
+      "step": 3910
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.163732519863892e-05,
+      "loss": 1.271,
+      "step": 3915
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.157094294407756e-05,
+      "loss": 1.2551,
+      "step": 3920
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.150451393969364e-05,
+      "loss": 1.2906,
+      "step": 3925
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.143803832945601e-05,
+      "loss": 1.2748,
+      "step": 3930
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.137151625743457e-05,
+      "loss": 1.2584,
+      "step": 3935
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.130494786779987e-05,
+      "loss": 1.287,
+      "step": 3940
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.123833330482284e-05,
+      "loss": 1.2745,
+      "step": 3945
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.117167271287453e-05,
+      "loss": 1.2545,
+      "step": 3950
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.110496623642568e-05,
+      "loss": 1.2647,
+      "step": 3955
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.103821402004654e-05,
+      "loss": 1.2756,
+      "step": 3960
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.097141620840647e-05,
+      "loss": 1.283,
+      "step": 3965
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.090457294627358e-05,
+      "loss": 1.2808,
+      "step": 3970
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.083768437851458e-05,
+      "loss": 1.2076,
+      "step": 3975
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.077075065009433e-05,
+      "loss": 1.2647,
+      "step": 3980
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.070377190607555e-05,
+      "loss": 1.2513,
+      "step": 3985
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.063674829161853e-05,
+      "loss": 1.2099,
+      "step": 3990
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.056967995198082e-05,
+      "loss": 1.284,
+      "step": 3995
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.050256703251688e-05,
+      "loss": 1.2988,
+      "step": 4000
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 10770,
+  "num_train_epochs": 1,
+  "save_steps": 1000,
+  "total_flos": 1.1217086630068224e+19,
+  "trial_name": null,
+  "trial_params": null
+}