diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,5586 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 5.996116504854369,
+  "eval_steps": 500,
+  "global_step": 4632,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999997696992664e-05,
+      "loss": 13.1952,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999971788208856e-05,
+      "loss": 11.8816,
+      "step": 10
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999917092181405e-05,
+      "loss": 9.584,
+      "step": 15
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9998336095401365e-05,
+      "loss": 6.3056,
+      "step": 20
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.999721341246355e-05,
+      "loss": 2.5137,
+      "step": 25
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9995802885928347e-05,
+      "loss": 0.9605,
+      "step": 30
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9994104532037997e-05,
+      "loss": 0.3271,
+      "step": 35
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9992118370349084e-05,
+      "loss": 0.1821,
+      "step": 40
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9989844423732325e-05,
+      "loss": 0.1988,
+      "step": 45
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9987282718372274e-05,
+      "loss": 0.1868,
+      "step": 50
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9984433283767034e-05,
+      "loss": 0.1084,
+      "step": 55
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.998129615272793e-05,
+      "loss": 0.0851,
+      "step": 60
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9977871361379114e-05,
+      "loss": 0.1002,
+      "step": 65
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.997415894915714e-05,
+      "loss": 0.1035,
+      "step": 70
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.997015895881054e-05,
+      "loss": 0.0681,
+      "step": 75
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9965871436399326e-05,
+      "loss": 0.0538,
+      "step": 80
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9961296431294426e-05,
+      "loss": 0.0531,
+      "step": 85
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9956433996177145e-05,
+      "loss": 0.0466,
+      "step": 90
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.995128418703856e-05,
+      "loss": 0.1114,
+      "step": 95
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.994584706317887e-05,
+      "loss": 0.088,
+      "step": 100
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.99401226872067e-05,
+      "loss": 0.0354,
+      "step": 105
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.993411112503839e-05,
+      "loss": 0.0422,
+      "step": 110
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.992781244589725e-05,
+      "loss": 0.0314,
+      "step": 115
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.992122672231275e-05,
+      "loss": 0.0959,
+      "step": 120
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.991435403011967e-05,
+      "loss": 0.0711,
+      "step": 125
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.990719444845724e-05,
+      "loss": 0.0692,
+      "step": 130
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.989974805976826e-05,
+      "loss": 0.0539,
+      "step": 135
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.989201494979808e-05,
+      "loss": 0.035,
+      "step": 140
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.988399520759368e-05,
+      "loss": 0.0523,
+      "step": 145
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9875688925502616e-05,
+      "loss": 0.0885,
+      "step": 150
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9867096199171955e-05,
+      "loss": 0.1449,
+      "step": 155
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9858217127547166e-05,
+      "loss": 0.0715,
+      "step": 160
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.984905181287101e-05,
+      "loss": 0.0659,
+      "step": 165
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9839600360682326e-05,
+      "loss": 0.0316,
+      "step": 170
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.982986287981486e-05,
+      "loss": 0.0829,
+      "step": 175
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.981983948239596e-05,
+      "loss": 0.1417,
+      "step": 180
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.980953028384533e-05,
+      "loss": 0.0533,
+      "step": 185
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.979893540287367e-05,
+      "loss": 0.0411,
+      "step": 190
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.978805496148132e-05,
+      "loss": 0.0272,
+      "step": 195
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.977688908495687e-05,
+      "loss": 0.0553,
+      "step": 200
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.976543790187569e-05,
+      "loss": 0.0486,
+      "step": 205
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.975370154409846e-05,
+      "loss": 0.0634,
+      "step": 210
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.974168014676967e-05,
+      "loss": 0.1162,
+      "step": 215
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.972937384831604e-05,
+      "loss": 0.0714,
+      "step": 220
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.971678279044493e-05,
+      "loss": 0.046,
+      "step": 225
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9703907118142715e-05,
+      "loss": 0.0458,
+      "step": 230
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.969074697967312e-05,
+      "loss": 0.0401,
+      "step": 235
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.96773025265755e-05,
+      "loss": 0.0512,
+      "step": 240
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.966357391366309e-05,
+      "loss": 0.0337,
+      "step": 245
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.964956129902126e-05,
+      "loss": 0.0669,
+      "step": 250
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9635264844005635e-05,
+      "loss": 0.0694,
+      "step": 255
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.96206847132403e-05,
+      "loss": 0.1863,
+      "step": 260
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9605821074615853e-05,
+      "loss": 0.0643,
+      "step": 265
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.95906740992875e-05,
+      "loss": 0.0181,
+      "step": 270
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.957524396167309e-05,
+      "loss": 0.0674,
+      "step": 275
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9559530839451055e-05,
+      "loss": 0.0956,
+      "step": 280
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.954353491355845e-05,
+      "loss": 0.0631,
+      "step": 285
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.952725636818879e-05,
+      "loss": 0.0688,
+      "step": 290
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.951069539078997e-05,
+      "loss": 0.0184,
+      "step": 295
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.949385217206211e-05,
+      "loss": 0.0465,
+      "step": 300
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.947672690595532e-05,
+      "loss": 0.0943,
+      "step": 305
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.945931978966752e-05,
+      "loss": 0.0526,
+      "step": 310
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.944163102364212e-05,
+      "loss": 0.0593,
+      "step": 315
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.9423660811565756e-05,
+      "loss": 0.0715,
+      "step": 320
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.940540936036591e-05,
+      "loss": 0.0666,
+      "step": 325
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.9386876880208545e-05,
+      "loss": 0.0725,
+      "step": 330
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.936806358449566e-05,
+      "loss": 0.0628,
+      "step": 335
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9348969689862894e-05,
+      "loss": 0.078,
+      "step": 340
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9329595416176956e-05,
+      "loss": 0.0824,
+      "step": 345
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9309940986533144e-05,
+      "loss": 0.0172,
+      "step": 350
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.9290006627252746e-05,
+      "loss": 0.1111,
+      "step": 355
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.926979256788046e-05,
+      "loss": 0.0104,
+      "step": 360
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.9249299041181726e-05,
+      "loss": 0.0147,
+      "step": 365
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.922852628314008e-05,
+      "loss": 0.0786,
+      "step": 370
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.920747453295439e-05,
+      "loss": 0.0298,
+      "step": 375
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.918614403303613e-05,
+      "loss": 0.0192,
+      "step": 380
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.916453502900661e-05,
+      "loss": 0.0632,
+      "step": 385
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.914264776969409e-05,
+      "loss": 0.0615,
+      "step": 390
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.912048250713096e-05,
+      "loss": 0.0617,
+      "step": 395
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9098039496550836e-05,
+      "loss": 0.0495,
+      "step": 400
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.907531899638559e-05,
+      "loss": 0.0369,
+      "step": 405
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9052321268262404e-05,
+      "loss": 0.0755,
+      "step": 410
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.9029046577000755e-05,
+      "loss": 0.0951,
+      "step": 415
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.900549519060934e-05,
+      "loss": 0.0368,
+      "step": 420
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.898166738028303e-05,
+      "loss": 0.0566,
+      "step": 425
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.895756342039971e-05,
+      "loss": 0.0586,
+      "step": 430
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.893318358851714e-05,
+      "loss": 0.0409,
+      "step": 435
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.890852816536974e-05,
+      "loss": 0.0133,
+      "step": 440
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.88835974348654e-05,
+      "loss": 0.0103,
+      "step": 445
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.885839168408215e-05,
+      "loss": 0.0702,
+      "step": 450
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.883291120326492e-05,
+      "loss": 0.032,
+      "step": 455
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.880715628582212e-05,
+      "loss": 0.0388,
+      "step": 460
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.878112722832235e-05,
+      "loss": 0.0147,
+      "step": 465
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.875482433049092e-05,
+      "loss": 0.0225,
+      "step": 470
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.87282478952064e-05,
+      "loss": 0.0239,
+      "step": 475
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.8701398228497196e-05,
+      "loss": 0.0235,
+      "step": 480
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.8674275639537936e-05,
+      "loss": 0.0652,
+      "step": 485
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.864688044064598e-05,
+      "loss": 0.0383,
+      "step": 490
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.86192129472778e-05,
+      "loss": 0.089,
+      "step": 495
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.859127347802534e-05,
+      "loss": 0.0361,
+      "step": 500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.856306235461234e-05,
+      "loss": 0.0502,
+      "step": 505
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.853457990189068e-05,
+      "loss": 0.0838,
+      "step": 510
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.850582644783657e-05,
+      "loss": 0.0329,
+      "step": 515
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.847680232354682e-05,
+      "loss": 0.0713,
+      "step": 520
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.844750786323503e-05,
+      "loss": 0.1027,
+      "step": 525
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.84179434042277e-05,
+      "loss": 0.0359,
+      "step": 530
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.8388109286960386e-05,
+      "loss": 0.0589,
+      "step": 535
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.835800585497377e-05,
+      "loss": 0.0383,
+      "step": 540
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.83276334549097e-05,
+      "loss": 0.0431,
+      "step": 545
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.829699243650719e-05,
+      "loss": 0.0564,
+      "step": 550
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.826608315259841e-05,
+      "loss": 0.015,
+      "step": 555
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.82349059591046e-05,
+      "loss": 0.0311,
+      "step": 560
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.8203461215032006e-05,
+      "loss": 0.0247,
+      "step": 565
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.8171749282467726e-05,
+      "loss": 0.0331,
+      "step": 570
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.8139770526575515e-05,
+      "loss": 0.0215,
+      "step": 575
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.810752531559164e-05,
+      "loss": 0.0534,
+      "step": 580
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.807501402082057e-05,
+      "loss": 0.0592,
+      "step": 585
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.804223701663078e-05,
+      "loss": 0.0591,
+      "step": 590
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.800919468045035e-05,
+      "loss": 0.0841,
+      "step": 595
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.79758873927627e-05,
+      "loss": 0.0856,
+      "step": 600
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.794231553710215e-05,
+      "loss": 0.031,
+      "step": 605
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.790847950004954e-05,
+      "loss": 0.0595,
+      "step": 610
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.7874379671227776e-05,
+      "loss": 0.0849,
+      "step": 615
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.78400164432973e-05,
+      "loss": 0.0424,
+      "step": 620
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.780539021195164e-05,
+      "loss": 0.0528,
+      "step": 625
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.77705013759128e-05,
+      "loss": 0.0399,
+      "step": 630
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.773535033692667e-05,
+      "loss": 0.0703,
+      "step": 635
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.7699937499758426e-05,
+      "loss": 0.0568,
+      "step": 640
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.7664263272187874e-05,
+      "loss": 0.0562,
+      "step": 645
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.762832806500471e-05,
+      "loss": 0.08,
+      "step": 650
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.759213229200384e-05,
+      "loss": 0.0535,
+      "step": 655
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.755567636998057e-05,
+      "loss": 0.0138,
+      "step": 660
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7518960718725854e-05,
+      "loss": 0.0132,
+      "step": 665
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.748198576102141e-05,
+      "loss": 0.028,
+      "step": 670
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.744475192263489e-05,
+      "loss": 0.1027,
+      "step": 675
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.740725963231497e-05,
+      "loss": 0.0592,
+      "step": 680
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.736950932178639e-05,
+      "loss": 0.0488,
+      "step": 685
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.7331501425745026e-05,
+      "loss": 0.0337,
+      "step": 690
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7293236381852835e-05,
+      "loss": 0.0141,
+      "step": 695
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.7254714630732844e-05,
+      "loss": 0.0995,
+      "step": 700
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.721593661596407e-05,
+      "loss": 0.0921,
+      "step": 705
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.7176902784076415e-05,
+      "loss": 0.1098,
+      "step": 710
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.713761358454554e-05,
+      "loss": 0.0181,
+      "step": 715
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.7098069469787624e-05,
+      "loss": 0.0245,
+      "step": 720
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.705827089515426e-05,
+      "loss": 0.0321,
+      "step": 725
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.7018218318927114e-05,
+      "loss": 0.0367,
+      "step": 730
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.69779122023127e-05,
+      "loss": 0.0708,
+      "step": 735
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6937353009437045e-05,
+      "loss": 0.0172,
+      "step": 740
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6896541207340383e-05,
+      "loss": 0.0242,
+      "step": 745
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.685547726597172e-05,
+      "loss": 0.038,
+      "step": 750
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.6814161658183485e-05,
+      "loss": 0.0576,
+      "step": 755
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.677259485972602e-05,
+      "loss": 0.0457,
+      "step": 760
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.673077734924214e-05,
+      "loss": 0.0421,
+      "step": 765
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.668870960826164e-05,
+      "loss": 0.0246,
+      "step": 770
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.664639212119569e-05,
+      "loss": 0.0663,
+      "step": 775
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.660382537533134e-05,
+      "loss": 0.029,
+      "step": 780
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.656100986082582e-05,
+      "loss": 0.0414,
+      "step": 785
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.6517946070700956e-05,
+      "loss": 0.0446,
+      "step": 790
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.6474634500837466e-05,
+      "loss": 0.107,
+      "step": 795
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.643107564996927e-05,
+      "loss": 0.0463,
+      "step": 800
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.638727001967772e-05,
+      "loss": 0.0542,
+      "step": 805
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.634321811438585e-05,
+      "loss": 0.0438,
+      "step": 810
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.6298920441352575e-05,
+      "loss": 0.0602,
+      "step": 815
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.6254377510666805e-05,
+      "loss": 0.0266,
+      "step": 820
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.6209589835241593e-05,
+      "loss": 0.0485,
+      "step": 825
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.6164557930808264e-05,
+      "loss": 0.0437,
+      "step": 830
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.6119282315910426e-05,
+      "loss": 0.0731,
+      "step": 835
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.6073763511898024e-05,
+      "loss": 0.0038,
+      "step": 840
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.6028002042921334e-05,
+      "loss": 0.0128,
+      "step": 845
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.598199843592493e-05,
+      "loss": 0.0477,
+      "step": 850
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.5935753220641606e-05,
+      "loss": 0.0358,
+      "step": 855
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.5889266929586284e-05,
+      "loss": 0.0441,
+      "step": 860
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.584254009804987e-05,
+      "loss": 0.0955,
+      "step": 865
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.579557326409311e-05,
+      "loss": 0.0148,
+      "step": 870
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.57483669685404e-05,
+      "loss": 0.0319,
+      "step": 875
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.570092175497351e-05,
+      "loss": 0.0227,
+      "step": 880
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.565323816972538e-05,
+      "loss": 0.0418,
+      "step": 885
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.5605316761873805e-05,
+      "loss": 0.0682,
+      "step": 890
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.5557158083235105e-05,
+      "loss": 0.0381,
+      "step": 895
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.550876268835779e-05,
+      "loss": 0.0327,
+      "step": 900
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.546013113451614e-05,
+      "loss": 0.0718,
+      "step": 905
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.541126398170385e-05,
+      "loss": 0.0239,
+      "step": 910
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.5362161792627524e-05,
+      "loss": 0.0271,
+      "step": 915
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.53128251327002e-05,
+      "loss": 0.0462,
+      "step": 920
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.5263254570034876e-05,
+      "loss": 0.0302,
+      "step": 925
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.521345067543794e-05,
+      "loss": 0.0351,
+      "step": 930
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.51634140224026e-05,
+      "loss": 0.0046,
+      "step": 935
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.51131451871023e-05,
+      "loss": 0.1034,
+      "step": 940
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.506264474838403e-05,
+      "loss": 0.037,
+      "step": 945
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.5011913287761755e-05,
+      "loss": 0.0634,
+      "step": 950
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.496095138940962e-05,
+      "loss": 0.0407,
+      "step": 955
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.490975964015527e-05,
+      "loss": 0.0154,
+      "step": 960
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.485833862947312e-05,
+      "loss": 0.0086,
+      "step": 965
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.48066889494775e-05,
+      "loss": 0.0421,
+      "step": 970
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.475481119491591e-05,
+      "loss": 0.0058,
+      "step": 975
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.470270596316211e-05,
+      "loss": 0.0564,
+      "step": 980
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.465037385420925e-05,
+      "loss": 0.0075,
+      "step": 985
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.4597815470663015e-05,
+      "loss": 0.0805,
+      "step": 990
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.454503141773464e-05,
+      "loss": 0.0223,
+      "step": 995
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.4492022303233926e-05,
+      "loss": 0.0668,
+      "step": 1000
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.4438788737562286e-05,
+      "loss": 0.019,
+      "step": 1005
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.4385331333705694e-05,
+      "loss": 0.0108,
+      "step": 1010
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.43316507072276e-05,
+      "loss": 0.0091,
+      "step": 1015
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.427774747626191e-05,
+      "loss": 0.0276,
+      "step": 1020
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.422362226150579e-05,
+      "loss": 0.0175,
+      "step": 1025
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.4169275686212584e-05,
+      "loss": 0.0793,
+      "step": 1030
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.4114708376184586e-05,
+      "loss": 0.0752,
+      "step": 1035
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.4059920959765866e-05,
+      "loss": 0.0273,
+      "step": 1040
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.4004914067835034e-05,
+      "loss": 0.0158,
+      "step": 1045
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.3949688333797955e-05,
+      "loss": 0.0169,
+      "step": 1050
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.3894244393580463e-05,
+      "loss": 0.0104,
+      "step": 1055
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.383858288562105e-05,
+      "loss": 0.0249,
+      "step": 1060
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.3782704450863516e-05,
+      "loss": 0.0582,
+      "step": 1065
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.3726609732749555e-05,
+      "loss": 0.1039,
+      "step": 1070
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.3670299377211396e-05,
+      "loss": 0.0135,
+      "step": 1075
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.361377403266431e-05,
+      "loss": 0.0334,
+      "step": 1080
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.355703434999921e-05,
+      "loss": 0.0358,
+      "step": 1085
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.350008098257508e-05,
+      "loss": 0.0033,
+      "step": 1090
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.34429145862115e-05,
+      "loss": 0.0216,
+      "step": 1095
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.338553581918111e-05,
+      "loss": 0.036,
+      "step": 1100
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.3327945342201973e-05,
+      "loss": 0.0328,
+      "step": 1105
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.327014381843002e-05,
+      "loss": 0.0261,
+      "step": 1110
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.321213191345136e-05,
+      "loss": 0.0531,
+      "step": 1115
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.315391029527469e-05,
+      "loss": 0.0252,
+      "step": 1120
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.309547963432353e-05,
+      "loss": 0.0398,
+      "step": 1125
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.303684060342854e-05,
+      "loss": 0.0358,
+      "step": 1130
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.297799387781977e-05,
+      "loss": 0.0085,
+      "step": 1135
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.291894013511888e-05,
+      "loss": 0.0163,
+      "step": 1140
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.285968005533133e-05,
+      "loss": 0.0154,
+      "step": 1145
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.280021432083857e-05,
+      "loss": 0.0399,
+      "step": 1150
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.274054361639015e-05,
+      "loss": 0.0089,
+      "step": 1155
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.268066862909588e-05,
+      "loss": 0.0367,
+      "step": 1160
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.2620590048417876e-05,
+      "loss": 0.0292,
+      "step": 1165
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.256030856616263e-05,
+      "loss": 0.0332,
+      "step": 1170
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.249982487647308e-05,
+      "loss": 0.0429,
+      "step": 1175
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.2439139675820556e-05,
+      "loss": 0.0393,
+      "step": 1180
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.237825366299682e-05,
+      "loss": 0.0249,
+      "step": 1185
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.231716753910595e-05,
+      "loss": 0.0399,
+      "step": 1190
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.225588200755637e-05,
+      "loss": 0.0159,
+      "step": 1195
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.219439777405264e-05,
+      "loss": 0.0887,
+      "step": 1200
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.2132715546587384e-05,
+      "loss": 0.0185,
+      "step": 1205
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.207083603543316e-05,
+      "loss": 0.0233,
+      "step": 1210
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.200875995313422e-05,
+      "loss": 0.0298,
+      "step": 1215
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.194648801449837e-05,
+      "loss": 0.0345,
+      "step": 1220
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.1884020936588666e-05,
+      "loss": 0.0081,
+      "step": 1225
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.182135943871524e-05,
+      "loss": 0.0061,
+      "step": 1230
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.175850424242695e-05,
+      "loss": 0.0084,
+      "step": 1235
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.16954560715031e-05,
+      "loss": 0.0609,
+      "step": 1240
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.163221565194512e-05,
+      "loss": 0.0488,
+      "step": 1245
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.1568783711968164e-05,
+      "loss": 0.0121,
+      "step": 1250
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.1505160981992754e-05,
+      "loss": 0.0338,
+      "step": 1255
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.144134819463638e-05,
+      "loss": 0.0139,
+      "step": 1260
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.137734608470502e-05,
+      "loss": 0.0051,
+      "step": 1265
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.1313155389184744e-05,
+      "loss": 0.0181,
+      "step": 1270
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.1248776847233164e-05,
+      "loss": 0.0801,
+      "step": 1275
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.118421120017095e-05,
+      "loss": 0.0032,
+      "step": 1280
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.111945919147329e-05,
+      "loss": 0.0027,
+      "step": 1285
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.105452156676135e-05,
+      "loss": 0.0342,
+      "step": 1290
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.098939907379366e-05,
+      "loss": 0.0031,
+      "step": 1295
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.092409246245751e-05,
+      "loss": 0.0112,
+      "step": 1300
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.08586024847603e-05,
+      "loss": 0.0077,
+      "step": 1305
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.079292989482094e-05,
+      "loss": 0.0258,
+      "step": 1310
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.0727075448861106e-05,
+      "loss": 0.0369,
+      "step": 1315
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.066103990519653e-05,
+      "loss": 0.0612,
+      "step": 1320
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.0594824024228305e-05,
+      "loss": 0.0672,
+      "step": 1325
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.052842856843413e-05,
+      "loss": 0.0389,
+      "step": 1330
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.046185430235949e-05,
+      "loss": 0.0722,
+      "step": 1335
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.0395101992608886e-05,
+      "loss": 0.0618,
+      "step": 1340
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.0328172407836994e-05,
+      "loss": 0.0159,
+      "step": 1345
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.026106631873981e-05,
+      "loss": 0.0319,
+      "step": 1350
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.01937844980458e-05,
+      "loss": 0.0569,
+      "step": 1355
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.012632772050697e-05,
+      "loss": 0.0152,
+      "step": 1360
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.005869676288995e-05,
+      "loss": 0.0044,
+      "step": 1365
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.999089240396708e-05,
+      "loss": 0.0242,
+      "step": 1370
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.9922915424507405e-05,
+      "loss": 0.0047,
+      "step": 1375
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.985476660726772e-05,
+      "loss": 0.0223,
+      "step": 1380
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.978644673698351e-05,
+      "loss": 0.0418,
+      "step": 1385
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.971795660035995e-05,
+      "loss": 0.0413,
+      "step": 1390
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.964929698606286e-05,
+      "loss": 0.0104,
+      "step": 1395
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.958046868470955e-05,
+      "loss": 0.0483,
+      "step": 1400
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.951147248885983e-05,
+      "loss": 0.0112,
+      "step": 1405
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.9442309193006755e-05,
+      "loss": 0.0254,
+      "step": 1410
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.937297959356759e-05,
+      "loss": 0.022,
+      "step": 1415
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.930348448887456e-05,
+      "loss": 0.0479,
+      "step": 1420
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.923382467916572e-05,
+      "loss": 0.0101,
+      "step": 1425
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.9164000966575676e-05,
+      "loss": 0.0398,
+      "step": 1430
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.909401415512638e-05,
+      "loss": 0.018,
+      "step": 1435
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.90238650507179e-05,
+      "loss": 0.0133,
+      "step": 1440
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.895355446111909e-05,
+      "loss": 0.0756,
+      "step": 1445
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.8883083195958315e-05,
+      "loss": 0.0379,
+      "step": 1450
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.881245206671412e-05,
+      "loss": 0.0038,
+      "step": 1455
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.8741661886705906e-05,
+      "loss": 0.0187,
+      "step": 1460
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.867071347108453e-05,
+      "loss": 0.0675,
+      "step": 1465
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.859960763682293e-05,
+      "loss": 0.0033,
+      "step": 1470
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.852834520270675e-05,
+      "loss": 0.0047,
+      "step": 1475
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.845692698932487e-05,
+      "loss": 0.0137,
+      "step": 1480
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.8385353819059974e-05,
+      "loss": 0.039,
+      "step": 1485
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.831362651607907e-05,
+      "loss": 0.0693,
+      "step": 1490
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.824174590632402e-05,
+      "loss": 0.0221,
+      "step": 1495
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.816971281750203e-05,
+      "loss": 0.045,
+      "step": 1500
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.809752807907607e-05,
+      "loss": 0.0121,
+      "step": 1505
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.80251925222554e-05,
+      "loss": 0.0035,
+      "step": 1510
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.7952706979985925e-05,
+      "loss": 0.0205,
+      "step": 1515
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.788007228694066e-05,
+      "loss": 0.0371,
+      "step": 1520
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.7807289279510074e-05,
+      "loss": 0.0988,
+      "step": 1525
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.77343587957925e-05,
+      "loss": 0.0574,
+      "step": 1530
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.766128167558446e-05,
+      "loss": 0.0045,
+      "step": 1535
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.7588058760370986e-05,
+      "loss": 0.0151,
+      "step": 1540
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.7514690893315953e-05,
+      "loss": 0.0572,
+      "step": 1545
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.7441178919252375e-05,
+      "loss": 0.0078,
+      "step": 1550
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.7367523684672636e-05,
+      "loss": 0.0659,
+      "step": 1555
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.729372603771878e-05,
+      "loss": 0.0188,
+      "step": 1560
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.7219786828172736e-05,
+      "loss": 0.0218,
+      "step": 1565
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.714570690744653e-05,
+      "loss": 0.021,
+      "step": 1570
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.7071487128572476e-05,
+      "loss": 0.0271,
+      "step": 1575
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.699712834619337e-05,
+      "loss": 0.0143,
+      "step": 1580
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.6922631416552625e-05,
+      "loss": 0.035,
+      "step": 1585
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.6847997197484406e-05,
+      "loss": 0.0257,
+      "step": 1590
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.677322654840382e-05,
+      "loss": 0.0037,
+      "step": 1595
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.669832033029693e-05,
+      "loss": 0.0152,
+      "step": 1600
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.6623279405710873e-05,
+      "loss": 0.0417,
+      "step": 1605
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.654810463874398e-05,
+      "loss": 0.0617,
+      "step": 1610
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.647279689503574e-05,
+      "loss": 0.0172,
+      "step": 1615
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.639735704175688e-05,
+      "loss": 0.0019,
+      "step": 1620
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.632178594759939e-05,
+      "loss": 0.029,
+      "step": 1625
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.624608448276647e-05,
+      "loss": 0.0111,
+      "step": 1630
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.6170253518962585e-05,
+      "loss": 0.0259,
+      "step": 1635
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.6094293929383346e-05,
+      "loss": 0.0027,
+      "step": 1640
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.601820658870551e-05,
+      "loss": 0.0019,
+      "step": 1645
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.5941992373076895e-05,
+      "loss": 0.0116,
+      "step": 1650
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.586565216010628e-05,
+      "loss": 0.0536,
+      "step": 1655
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.578918682885331e-05,
+      "loss": 0.0239,
+      "step": 1660
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.571259725981837e-05,
+      "loss": 0.0433,
+      "step": 1665
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.5635884334932445e-05,
+      "loss": 0.0147,
+      "step": 1670
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.5559048937546963e-05,
+      "loss": 0.0301,
+      "step": 1675
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.548209195242364e-05,
+      "loss": 0.0284,
+      "step": 1680
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.540501426572426e-05,
+      "loss": 0.0218,
+      "step": 1685
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.53278167650005e-05,
+      "loss": 0.0075,
+      "step": 1690
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5250500339183696e-05,
+      "loss": 0.0128,
+      "step": 1695
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5173065878574606e-05,
+      "loss": 0.0248,
+      "step": 1700
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.509551427483318e-05,
+      "loss": 0.01,
+      "step": 1705
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.5017846420968234e-05,
+      "loss": 0.0134,
+      "step": 1710
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.494006321132726e-05,
+      "loss": 0.014,
+      "step": 1715
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.4862165541586026e-05,
+      "loss": 0.0233,
+      "step": 1720
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.478415430873836e-05,
+      "loss": 0.054,
+      "step": 1725
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.470603041108572e-05,
+      "loss": 0.0387,
+      "step": 1730
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.462779474822695e-05,
+      "loss": 0.0535,
+      "step": 1735
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.454944822104784e-05,
+      "loss": 0.0014,
+      "step": 1740
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.447099173171079e-05,
+      "loss": 0.0417,
+      "step": 1745
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.439242618364443e-05,
+      "loss": 0.0511,
+      "step": 1750
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.431375248153321e-05,
+      "loss": 0.0474,
+      "step": 1755
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.423497153130695e-05,
+      "loss": 0.0084,
+      "step": 1760
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.4156084240130465e-05,
+      "loss": 0.0165,
+      "step": 1765
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.4077091516393055e-05,
+      "loss": 0.0385,
+      "step": 1770
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.399799426969812e-05,
+      "loss": 0.0192,
+      "step": 1775
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.391879341085261e-05,
+      "loss": 0.004,
+      "step": 1780
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.3839489851856587e-05,
+      "loss": 0.0235,
+      "step": 1785
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.3760084505892717e-05,
+      "loss": 0.0221,
+      "step": 1790
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.3680578287315714e-05,
+      "loss": 0.0177,
+      "step": 1795
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.360097211164188e-05,
+      "loss": 0.003,
+      "step": 1800
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.352126689553849e-05,
+      "loss": 0.0583,
+      "step": 1805
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.3441463556813324e-05,
+      "loss": 0.0081,
+      "step": 1810
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.336156301440398e-05,
+      "loss": 0.0029,
+      "step": 1815
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.328156618836741e-05,
+      "loss": 0.0157,
+      "step": 1820
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.3201473999869265e-05,
+      "loss": 0.0273,
+      "step": 1825
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.312128737117327e-05,
+      "loss": 0.0093,
+      "step": 1830
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.304100722563067e-05,
+      "loss": 0.0017,
+      "step": 1835
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.296063448766953e-05,
+      "loss": 0.0245,
+      "step": 1840
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.288017008278414e-05,
+      "loss": 0.0555,
+      "step": 1845
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.279961493752431e-05,
+      "loss": 0.0284,
+      "step": 1850
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.271896997948476e-05,
+      "loss": 0.0397,
+      "step": 1855
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.263823613729439e-05,
+      "loss": 0.0628,
+      "step": 1860
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.2557414340605595e-05,
+      "loss": 0.0103,
+      "step": 1865
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.247650552008359e-05,
+      "loss": 0.0215,
+      "step": 1870
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.239551060739565e-05,
+      "loss": 0.037,
+      "step": 1875
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.231443053520043e-05,
+      "loss": 0.0017,
+      "step": 1880
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.223326623713716e-05,
+      "loss": 0.0138,
+      "step": 1885
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.215201864781498e-05,
+      "loss": 0.0016,
+      "step": 1890
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.20706887028021e-05,
+      "loss": 0.0175,
+      "step": 1895
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.1989277338615054e-05,
+      "loss": 0.0045,
+      "step": 1900
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.190778549270794e-05,
+      "loss": 0.0327,
+      "step": 1905
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.182621410346159e-05,
+      "loss": 0.0025,
+      "step": 1910
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.17445641101728e-05,
+      "loss": 0.0953,
+      "step": 1915
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.166283645304347e-05,
+      "loss": 0.0114,
+      "step": 1920
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.158103207316981e-05,
+      "loss": 0.0445,
+      "step": 1925
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.14991519125315e-05,
+      "loss": 0.0462,
+      "step": 1930
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.141719691398083e-05,
+      "loss": 0.0016,
+      "step": 1935
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.133516802123187e-05,
+      "loss": 0.0046,
+      "step": 1940
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.125306617884954e-05,
+      "loss": 0.0234,
+      "step": 1945
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.117089233223882e-05,
+      "loss": 0.0272,
+      "step": 1950
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.108864742763381e-05,
+      "loss": 0.0456,
+      "step": 1955
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.1006332412086824e-05,
+      "loss": 0.0057,
+      "step": 1960
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.092394823345754e-05,
+      "loss": 0.0214,
+      "step": 1965
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.084149584040203e-05,
+      "loss": 0.0047,
+      "step": 1970
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.0758976182361846e-05,
+      "loss": 0.0336,
+      "step": 1975
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.0676390209553114e-05,
+      "loss": 0.0531,
+      "step": 1980
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.0593738872955586e-05,
+      "loss": 0.0075,
+      "step": 1985
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.0511023124301664e-05,
+      "loss": 0.0021,
+      "step": 1990
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.0428243916065447e-05,
+      "loss": 0.0083,
+      "step": 1995
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.03454022014518e-05,
+      "loss": 0.0076,
+      "step": 2000
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.0262498934385336e-05,
+      "loss": 0.0041,
+      "step": 2005
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.0179535069499443e-05,
+      "loss": 0.0007,
+      "step": 2010
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.00965115621253e-05,
+      "loss": 0.0024,
+      "step": 2015
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.0013429368280865e-05,
+      "loss": 0.0017,
+      "step": 2020
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.993028944465988e-05,
+      "loss": 0.0112,
+      "step": 2025
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.9847092748620843e-05,
+      "loss": 0.0224,
+      "step": 2030
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.976384023817599e-05,
+      "loss": 0.0046,
+      "step": 2035
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.968053287198026e-05,
+      "loss": 0.0022,
+      "step": 2040
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.9597171609320258e-05,
+      "loss": 0.0007,
+      "step": 2045
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.9513757410103203e-05,
+      "loss": 0.0576,
+      "step": 2050
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.9430291234845896e-05,
+      "loss": 0.0401,
+      "step": 2055
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.934677404466362e-05,
+      "loss": 0.0033,
+      "step": 2060
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.9263206801259108e-05,
+      "loss": 0.0412,
+      "step": 2065
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.9179590466911466e-05,
+      "loss": 0.0398,
+      "step": 2070
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.9095926004465068e-05,
+      "loss": 0.0112,
+      "step": 2075
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.9012214377318502e-05,
+      "loss": 0.0131,
+      "step": 2080
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.892845654941344e-05,
+      "loss": 0.1122,
+      "step": 2085
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.884465348522356e-05,
+      "loss": 0.0268,
+      "step": 2090
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.876080614974346e-05,
+      "loss": 0.006,
+      "step": 2095
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.8676915508477487e-05,
+      "loss": 0.0191,
+      "step": 2100
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.8592982527428686e-05,
+      "loss": 0.03,
+      "step": 2105
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.8509008173087625e-05,
+      "loss": 0.0169,
+      "step": 2110
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.8424993412421308e-05,
+      "loss": 0.0039,
+      "step": 2115
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.8340939212861994e-05,
+      "loss": 0.0163,
+      "step": 2120
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.825684654229609e-05,
+      "loss": 0.0336,
+      "step": 2125
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.817271636905302e-05,
+      "loss": 0.0213,
+      "step": 2130
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.8088549661894016e-05,
+      "loss": 0.0525,
+      "step": 2135
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.8004347390001034e-05,
+      "loss": 0.0067,
+      "step": 2140
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7920110522965532e-05,
+      "loss": 0.0027,
+      "step": 2145
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7835840030777333e-05,
+      "loss": 0.0337,
+      "step": 2150
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.7751536883813485e-05,
+      "loss": 0.0275,
+      "step": 2155
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.7667202052827023e-05,
+      "loss": 0.0051,
+      "step": 2160
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.7582836508935838e-05,
+      "loss": 0.0194,
+      "step": 2165
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.749844122361147e-05,
+      "loss": 0.023,
+      "step": 2170
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.7414017168667965e-05,
+      "loss": 0.0056,
+      "step": 2175
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.732956531625061e-05,
+      "loss": 0.0401,
+      "step": 2180
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.72450866388248e-05,
+      "loss": 0.0005,
+      "step": 2185
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.7160582109164833e-05,
+      "loss": 0.0016,
+      "step": 2190
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.7076052700342675e-05,
+      "loss": 0.0359,
+      "step": 2195
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.6991499385716783e-05,
+      "loss": 0.0854,
+      "step": 2200
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.69069231389209e-05,
+      "loss": 0.0253,
+      "step": 2205
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.6822324933852816e-05,
+      "loss": 0.0239,
+      "step": 2210
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.6737705744663188e-05,
+      "loss": 0.0133,
+      "step": 2215
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.6653066545744298e-05,
+      "loss": 0.0209,
+      "step": 2220
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.6568408311718834e-05,
+      "loss": 0.0436,
+      "step": 2225
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.648373201742868e-05,
+      "loss": 0.0127,
+      "step": 2230
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.6399038637923694e-05,
+      "loss": 0.0366,
+      "step": 2235
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.6314329148450455e-05,
+      "loss": 0.006,
+      "step": 2240
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.622960452444106e-05,
+      "loss": 0.0302,
+      "step": 2245
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.6144865741501873e-05,
+      "loss": 0.0429,
+      "step": 2250
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.6060113775402295e-05,
+      "loss": 0.0072,
+      "step": 2255
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.5975349602063554e-05,
+      "loss": 0.0077,
+      "step": 2260
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.5890574197547414e-05,
+      "loss": 0.0255,
+      "step": 2265
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.5805788538044988e-05,
+      "loss": 0.0196,
+      "step": 2270
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.572099359986547e-05,
+      "loss": 0.0321,
+      "step": 2275
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.563619035942489e-05,
+      "loss": 0.0332,
+      "step": 2280
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.5551379793234886e-05,
+      "loss": 0.0081,
+      "step": 2285
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.546656287789146e-05,
+      "loss": 0.0157,
+      "step": 2290
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.5381740590063708e-05,
+      "loss": 0.037,
+      "step": 2295
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.5296913906482596e-05,
+      "loss": 0.0109,
+      "step": 2300
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.521208380392972e-05,
+      "loss": 0.0565,
+      "step": 2305
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.5127251259226036e-05,
+      "loss": 0.0107,
+      "step": 2310
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.5042417249220606e-05,
+      "loss": 0.0012,
+      "step": 2315
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.49575827507794e-05,
+      "loss": 0.0022,
+      "step": 2320
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.4872748740773976e-05,
+      "loss": 0.0218,
+      "step": 2325
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.4787916196070276e-05,
+      "loss": 0.0465,
+      "step": 2330
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.4703086093517403e-05,
+      "loss": 0.0306,
+      "step": 2335
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.4618259409936298e-05,
+      "loss": 0.0403,
+      "step": 2340
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.453343712210855e-05,
+      "loss": 0.0162,
+      "step": 2345
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.4448620206765123e-05,
+      "loss": 0.0056,
+      "step": 2350
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.436380964057511e-05,
+      "loss": 0.0078,
+      "step": 2355
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.4279006400134535e-05,
+      "loss": 0.0059,
+      "step": 2360
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.4194211461955015e-05,
+      "loss": 0.0039,
+      "step": 2365
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.410942580245259e-05,
+      "loss": 0.0016,
+      "step": 2370
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.4024650397936455e-05,
+      "loss": 0.0004,
+      "step": 2375
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.3939886224597704e-05,
+      "loss": 0.0415,
+      "step": 2380
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.3855134258498136e-05,
+      "loss": 0.0329,
+      "step": 2385
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.377039547555895e-05,
+      "loss": 0.0137,
+      "step": 2390
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.368567085154955e-05,
+      "loss": 0.0118,
+      "step": 2395
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2.3600961362076315e-05,
+      "loss": 0.0038,
+      "step": 2400
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2.3516267982571318e-05,
+      "loss": 0.009,
+      "step": 2405
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.343159168828117e-05,
+      "loss": 0.001,
+      "step": 2410
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.3346933454255708e-05,
+      "loss": 0.0035,
+      "step": 2415
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.3262294255336814e-05,
+      "loss": 0.0026,
+      "step": 2420
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.317767506614718e-05,
+      "loss": 0.0113,
+      "step": 2425
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.3093076861079103e-05,
+      "loss": 0.0244,
+      "step": 2430
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.300850061428322e-05,
+      "loss": 0.0041,
+      "step": 2435
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2.292394729965733e-05,
+      "loss": 0.0715,
+      "step": 2440
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.2839417890835172e-05,
+      "loss": 0.0036,
+      "step": 2445
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.2754913361175198e-05,
+      "loss": 0.0101,
+      "step": 2450
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.2670434683749395e-05,
+      "loss": 0.0025,
+      "step": 2455
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.258598283133204e-05,
+      "loss": 0.0029,
+      "step": 2460
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2.2501558776388533e-05,
+      "loss": 0.0454,
+      "step": 2465
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2.2417163491064175e-05,
+      "loss": 0.0045,
+      "step": 2470
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2.2332797947172983e-05,
+      "loss": 0.0028,
+      "step": 2475
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2.224846311618652e-05,
+      "loss": 0.0031,
+      "step": 2480
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2.216415996922267e-05,
+      "loss": 0.0229,
+      "step": 2485
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2.2079889477034484e-05,
+      "loss": 0.0127,
+      "step": 2490
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2.199565260999898e-05,
+      "loss": 0.0076,
+      "step": 2495
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2.1911450338105983e-05,
+      "loss": 0.0094,
+      "step": 2500
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2.1827283630946986e-05,
+      "loss": 0.0175,
+      "step": 2505
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2.1743153457703913e-05,
+      "loss": 0.0019,
+      "step": 2510
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2.165906078713802e-05,
+      "loss": 0.0016,
+      "step": 2515
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2.15750065875787e-05,
+      "loss": 0.0423,
+      "step": 2520
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2.1490991826912378e-05,
+      "loss": 0.0102,
+      "step": 2525
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2.1407017472571323e-05,
+      "loss": 0.0244,
+      "step": 2530
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2.1323084491522522e-05,
+      "loss": 0.0082,
+      "step": 2535
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 2.1239193850256554e-05,
+      "loss": 0.0019,
+      "step": 2540
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 2.1155346514776437e-05,
+      "loss": 0.0265,
+      "step": 2545
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 2.1071543450586563e-05,
+      "loss": 0.0029,
+      "step": 2550
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 2.0987785622681504e-05,
+      "loss": 0.0012,
+      "step": 2555
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 2.0904073995534934e-05,
+      "loss": 0.0161,
+      "step": 2560
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 2.0820409533088543e-05,
+      "loss": 0.0243,
+      "step": 2565
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 2.073679319874089e-05,
+      "loss": 0.0103,
+      "step": 2570
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 2.0653225955336385e-05,
+      "loss": 0.0045,
+      "step": 2575
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 2.056970876515411e-05,
+      "loss": 0.0036,
+      "step": 2580
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 2.04862425898968e-05,
+      "loss": 0.0027,
+      "step": 2585
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 2.0402828390679748e-05,
+      "loss": 0.0004,
+      "step": 2590
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 2.0319467128019744e-05,
+      "loss": 0.0098,
+      "step": 2595
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 2.0236159761824013e-05,
+      "loss": 0.0021,
+      "step": 2600
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 2.0152907251379162e-05,
+      "loss": 0.0022,
+      "step": 2605
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 2.0069710555340133e-05,
+      "loss": 0.031,
+      "step": 2610
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.9986570631719138e-05,
+      "loss": 0.0006,
+      "step": 2615
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.9903488437874703e-05,
+      "loss": 0.0003,
+      "step": 2620
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.982046493050056e-05,
+      "loss": 0.053,
+      "step": 2625
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.973750106561467e-05,
+      "loss": 0.0002,
+      "step": 2630
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.9654597798548204e-05,
+      "loss": 0.0311,
+      "step": 2635
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.9571756083934556e-05,
+      "loss": 0.0009,
+      "step": 2640
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.9488976875698345e-05,
+      "loss": 0.002,
+      "step": 2645
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.940626112704442e-05,
+      "loss": 0.0022,
+      "step": 2650
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.932360979044689e-05,
+      "loss": 0.0032,
+      "step": 2655
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.9241023817638167e-05,
+      "loss": 0.0012,
+      "step": 2660
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.9158504159597976e-05,
+      "loss": 0.0005,
+      "step": 2665
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.907605176654246e-05,
+      "loss": 0.0053,
+      "step": 2670
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.899366758791318e-05,
+      "loss": 0.0105,
+      "step": 2675
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.89113525723662e-05,
+      "loss": 0.0003,
+      "step": 2680
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.882910766776119e-05,
+      "loss": 0.0004,
+      "step": 2685
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.874693382115047e-05,
+      "loss": 0.0273,
+      "step": 2690
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.8664831978768142e-05,
+      "loss": 0.0016,
+      "step": 2695
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.8582803086019173e-05,
+      "loss": 0.02,
+      "step": 2700
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.850084808746851e-05,
+      "loss": 0.0635,
+      "step": 2705
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.8418967926830194e-05,
+      "loss": 0.0084,
+      "step": 2710
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.8337163546956537e-05,
+      "loss": 0.005,
+      "step": 2715
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.8255435889827207e-05,
+      "loss": 0.0002,
+      "step": 2720
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.8173785896538416e-05,
+      "loss": 0.0395,
+      "step": 2725
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.8092214507292072e-05,
+      "loss": 0.0249,
+      "step": 2730
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.8010722661384955e-05,
+      "loss": 0.0274,
+      "step": 2735
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.792931129719791e-05,
+      "loss": 0.0322,
+      "step": 2740
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.784798135218502e-05,
+      "loss": 0.0151,
+      "step": 2745
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.776673376286284e-05,
+      "loss": 0.0061,
+      "step": 2750
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.768556946479958e-05,
+      "loss": 0.0237,
+      "step": 2755
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.7604489392604346e-05,
+      "loss": 0.0084,
+      "step": 2760
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.7523494479916413e-05,
+      "loss": 0.0008,
+      "step": 2765
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.7442585659394408e-05,
+      "loss": 0.009,
+      "step": 2770
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.736176386270562e-05,
+      "loss": 0.0055,
+      "step": 2775
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.7281030020515238e-05,
+      "loss": 0.0006,
+      "step": 2780
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.7200385062475688e-05,
+      "loss": 0.0024,
+      "step": 2785
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.7119829917215865e-05,
+      "loss": 0.0011,
+      "step": 2790
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.7039365512330473e-05,
+      "loss": 0.0145,
+      "step": 2795
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.6958992774369334e-05,
+      "loss": 0.0015,
+      "step": 2800
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.6878712628826727e-05,
+      "loss": 0.0523,
+      "step": 2805
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.6798526000130737e-05,
+      "loss": 0.0248,
+      "step": 2810
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.671843381163259e-05,
+      "loss": 0.0416,
+      "step": 2815
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.6638436985596023e-05,
+      "loss": 0.0291,
+      "step": 2820
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.655853644318669e-05,
+      "loss": 0.0016,
+      "step": 2825
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.6478733104461503e-05,
+      "loss": 0.002,
+      "step": 2830
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.6399027888358127e-05,
+      "loss": 0.0234,
+      "step": 2835
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.631942171268429e-05,
+      "loss": 0.0075,
+      "step": 2840
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.6239915494107293e-05,
+      "loss": 0.013,
+      "step": 2845
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.6160510148143416e-05,
+      "loss": 0.0151,
+      "step": 2850
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.608120658914739e-05,
+      "loss": 0.0044,
+      "step": 2855
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.6002005730301884e-05,
+      "loss": 0.0345,
+      "step": 2860
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.5922908483606947e-05,
+      "loss": 0.0012,
+      "step": 2865
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.5843915759869547e-05,
+      "loss": 0.0058,
+      "step": 2870
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.576502846869305e-05,
+      "loss": 0.0011,
+      "step": 2875
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.5686247518466794e-05,
+      "loss": 0.0372,
+      "step": 2880
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.560757381635557e-05,
+      "loss": 0.0005,
+      "step": 2885
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.5529008268289218e-05,
+      "loss": 0.0062,
+      "step": 2890
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.5450551778952173e-05,
+      "loss": 0.0034,
+      "step": 2895
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.5372205251773054e-05,
+      "loss": 0.013,
+      "step": 2900
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.529396958891428e-05,
+      "loss": 0.0016,
+      "step": 2905
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.521584569126165e-05,
+      "loss": 0.0204,
+      "step": 2910
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.513783445841398e-05,
+      "loss": 0.0006,
+      "step": 2915
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.5059936788672755e-05,
+      "loss": 0.0675,
+      "step": 2920
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.4982153579031772e-05,
+      "loss": 0.002,
+      "step": 2925
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.4904485725166834e-05,
+      "loss": 0.0049,
+      "step": 2930
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.48269341214254e-05,
+      "loss": 0.001,
+      "step": 2935
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.4749499660816315e-05,
+      "loss": 0.0051,
+      "step": 2940
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.4672183234999511e-05,
+      "loss": 0.0747,
+      "step": 2945
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.4594985734275742e-05,
+      "loss": 0.0197,
+      "step": 2950
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.4517908047576367e-05,
+      "loss": 0.021,
+      "step": 2955
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.4440951062453039e-05,
+      "loss": 0.0447,
+      "step": 2960
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.4364115665067566e-05,
+      "loss": 0.0152,
+      "step": 2965
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.4287402740181627e-05,
+      "loss": 0.0668,
+      "step": 2970
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.4210813171146692e-05,
+      "loss": 0.0014,
+      "step": 2975
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.4134347839893716e-05,
+      "loss": 0.0268,
+      "step": 2980
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.4058007626923109e-05,
+      "loss": 0.0307,
+      "step": 2985
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.3981793411294491e-05,
+      "loss": 0.0071,
+      "step": 2990
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.3905706070616655e-05,
+      "loss": 0.0026,
+      "step": 2995
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.3829746481037426e-05,
+      "loss": 0.0005,
+      "step": 3000
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.3753915517233532e-05,
+      "loss": 0.0346,
+      "step": 3005
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.3678214052400623e-05,
+      "loss": 0.035,
+      "step": 3010
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.3602642958243126e-05,
+      "loss": 0.002,
+      "step": 3015
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.3527203104964265e-05,
+      "loss": 0.0106,
+      "step": 3020
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.3451895361256017e-05,
+      "loss": 0.0348,
+      "step": 3025
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.3376720594289127e-05,
+      "loss": 0.0135,
+      "step": 3030
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.3301679669703079e-05,
+      "loss": 0.0009,
+      "step": 3035
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.3226773451596191e-05,
+      "loss": 0.0097,
+      "step": 3040
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.3152002802515599e-05,
+      "loss": 0.0027,
+      "step": 3045
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.3077368583447385e-05,
+      "loss": 0.0075,
+      "step": 3050
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.3002871653806637e-05,
+      "loss": 0.0864,
+      "step": 3055
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.2928512871427523e-05,
+      "loss": 0.0277,
+      "step": 3060
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.2854293092553465e-05,
+      "loss": 0.0479,
+      "step": 3065
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.278021317182726e-05,
+      "loss": 0.01,
+      "step": 3070
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.2706273962281223e-05,
+      "loss": 0.005,
+      "step": 3075
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.2632476315327365e-05,
+      "loss": 0.0124,
+      "step": 3080
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.2558821080747632e-05,
+      "loss": 0.0011,
+      "step": 3085
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.2500000000000006e-05,
+      "loss": 0.0228,
+      "step": 3090
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.2426603243903385e-05,
+      "loss": 0.0181,
+      "step": 3095
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.2353351270814936e-05,
+      "loss": 0.0058,
+      "step": 3100
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.2280244924233148e-05,
+      "loss": 0.0008,
+      "step": 3105
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.2207285045979586e-05,
+      "loss": 0.0166,
+      "step": 3110
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.2134472476189262e-05,
+      "loss": 0.0011,
+      "step": 3115
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.2061808053300889e-05,
+      "loss": 0.0022,
+      "step": 3120
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.1989292614047323e-05,
+      "loss": 0.0005,
+      "step": 3125
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.1916926993445806e-05,
+      "loss": 0.0042,
+      "step": 3130
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.1844712024788463e-05,
+      "loss": 0.0398,
+      "step": 3135
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.1772648539632616e-05,
+      "loss": 0.0007,
+      "step": 3140
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.1700737367791278e-05,
+      "loss": 0.0012,
+      "step": 3145
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.1628979337323579e-05,
+      "loss": 0.0176,
+      "step": 3150
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.1557375274525185e-05,
+      "loss": 0.0012,
+      "step": 3155
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.1485926003918855e-05,
+      "loss": 0.0017,
+      "step": 3160
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.1414632348244882e-05,
+      "loss": 0.0005,
+      "step": 3165
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.1343495128451678e-05,
+      "loss": 0.0026,
+      "step": 3170
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.1272515163686273e-05,
+      "loss": 0.0115,
+      "step": 3175
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.120169327128492e-05,
+      "loss": 0.0003,
+      "step": 3180
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.1131030266763629e-05,
+      "loss": 0.0008,
+      "step": 3185
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.106052696380884e-05,
+      "loss": 0.0151,
+      "step": 3190
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.0990184174268033e-05,
+      "loss": 0.0003,
+      "step": 3195
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.0920002708140336e-05,
+      "loss": 0.0038,
+      "step": 3200
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.0849983373567274e-05,
+      "loss": 0.0031,
+      "step": 3205
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.0780126976823385e-05,
+      "loss": 0.0006,
+      "step": 3210
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.0710434322306997e-05,
+      "loss": 0.0039,
+      "step": 3215
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.064090621253094e-05,
+      "loss": 0.0245,
+      "step": 3220
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.0571543448113314e-05,
+      "loss": 0.0007,
+      "step": 3225
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.050234682776823e-05,
+      "loss": 0.0013,
+      "step": 3230
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.0433317148296687e-05,
+      "loss": 0.0347,
+      "step": 3235
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.0364455204577309e-05,
+      "loss": 0.0004,
+      "step": 3240
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.0295761789557298e-05,
+      "loss": 0.0064,
+      "step": 3245
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.0227237694243188e-05,
+      "loss": 0.0024,
+      "step": 3250
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.015888370769183e-05,
+      "loss": 0.0142,
+      "step": 3255
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.0090700617001233e-05,
+      "loss": 0.0006,
+      "step": 3260
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.0022689207301564e-05,
+      "loss": 0.001,
+      "step": 3265
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 9.95485026174608e-06,
+      "loss": 0.0026,
+      "step": 3270
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 9.887184561502078e-06,
+      "loss": 0.0207,
+      "step": 3275
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 9.819692885741968e-06,
+      "loss": 0.004,
+      "step": 3280
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 9.75237601163422e-06,
+      "loss": 0.0092,
+      "step": 3285
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 9.68523471433452e-06,
+      "loss": 0.027,
+      "step": 3290
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 9.61826976697672e-06,
+      "loss": 0.0029,
+      "step": 3295
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 9.55148194066404e-06,
+      "loss": 0.0388,
+      "step": 3300
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 9.484872004460113e-06,
+      "loss": 0.0714,
+      "step": 3305
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 9.418440725380187e-06,
+      "loss": 0.0025,
+      "step": 3310
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 9.352188868382273e-06,
+      "loss": 0.0006,
+      "step": 3315
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 9.2861171963583e-06,
+      "loss": 0.0216,
+      "step": 3320
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 9.220226470125398e-06,
+      "loss": 0.0407,
+      "step": 3325
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.154517448417072e-06,
+      "loss": 0.0572,
+      "step": 3330
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.08899088787451e-06,
+      "loss": 0.0044,
+      "step": 3335
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.023647543037855e-06,
+      "loss": 0.0017,
+      "step": 3340
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 8.958488166337526e-06,
+      "loss": 0.0082,
+      "step": 3345
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.893513508085508e-06,
+      "loss": 0.0051,
+      "step": 3350
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.828724316466782e-06,
+      "loss": 0.0352,
+      "step": 3355
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.764121337530673e-06,
+      "loss": 0.0004,
+      "step": 3360
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.699705315182238e-06,
+      "loss": 0.0092,
+      "step": 3365
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.635476991173755e-06,
+      "loss": 0.0208,
+      "step": 3370
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.57143710509612e-06,
+      "loss": 0.0005,
+      "step": 3375
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.507586394370381e-06,
+      "loss": 0.0022,
+      "step": 3380
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.44392559423922e-06,
+      "loss": 0.0019,
+      "step": 3385
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 8.380455437758512e-06,
+      "loss": 0.0199,
+      "step": 3390
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 8.317176655788825e-06,
+      "loss": 0.0004,
+      "step": 3395
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 8.254089976987084e-06,
+      "loss": 0.0037,
+      "step": 3400
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 8.191196127798106e-06,
+      "loss": 0.0036,
+      "step": 3405
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 8.128495832446301e-06,
+      "loss": 0.0004,
+      "step": 3410
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 8.065989812927293e-06,
+      "loss": 0.0248,
+      "step": 3415
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 8.003678788999594e-06,
+      "loss": 0.0215,
+      "step": 3420
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 7.94156347817637e-06,
+      "loss": 0.0013,
+      "step": 3425
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 7.879644595717098e-06,
+      "loss": 0.0024,
+      "step": 3430
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 7.817922854619442e-06,
+      "loss": 0.0028,
+      "step": 3435
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 7.756398965610918e-06,
+      "loss": 0.0008,
+      "step": 3440
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 7.695073637140807e-06,
+      "loss": 0.0011,
+      "step": 3445
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 7.633947575371927e-06,
+      "loss": 0.0002,
+      "step": 3450
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 7.573021484172582e-06,
+      "loss": 0.005,
+      "step": 3455
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 7.51229606510836e-06,
+      "loss": 0.0159,
+      "step": 3460
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.451772017434142e-06,
+      "loss": 0.0263,
+      "step": 3465
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.391450038085976e-06,
+      "loss": 0.002,
+      "step": 3470
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.331330821673121e-06,
+      "loss": 0.0068,
+      "step": 3475
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.271415060470002e-06,
+      "loss": 0.0003,
+      "step": 3480
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.211703444408241e-06,
+      "loss": 0.0203,
+      "step": 3485
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 7.152196661068747e-06,
+      "loss": 0.0003,
+      "step": 3490
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 7.092895395673746e-06,
+      "loss": 0.009,
+      "step": 3495
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 7.033800331078938e-06,
+      "loss": 0.0026,
+      "step": 3500
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.974912147765614e-06,
+      "loss": 0.0003,
+      "step": 3505
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.916231523832828e-06,
+      "loss": 0.0227,
+      "step": 3510
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 6.857759134989552e-06,
+      "loss": 0.0109,
+      "step": 3515
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 6.799495654546967e-06,
+      "loss": 0.0006,
+      "step": 3520
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 6.741441753410627e-06,
+      "loss": 0.0056,
+      "step": 3525
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 6.683598100072805e-06,
+      "loss": 0.0079,
+      "step": 3530
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 6.625965360604758e-06,
+      "loss": 0.0006,
+      "step": 3535
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 6.568544198649046e-06,
+      "loss": 0.0039,
+      "step": 3540
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 6.511335275411934e-06,
+      "loss": 0.0008,
+      "step": 3545
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 6.454339249655739e-06,
+      "loss": 0.0273,
+      "step": 3550
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 6.397556777691263e-06,
+      "loss": 0.0023,
+      "step": 3555
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 6.340988513370216e-06,
+      "loss": 0.0013,
+      "step": 3560
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 6.284635108077724e-06,
+      "loss": 0.0586,
+      "step": 3565
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 6.228497210724782e-06,
+      "loss": 0.0182,
+      "step": 3570
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 6.1725754677408184e-06,
+      "loss": 0.0133,
+      "step": 3575
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 6.116870523066245e-06,
+      "loss": 0.0011,
+      "step": 3580
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.061383018145011e-06,
+      "loss": 0.0279,
+      "step": 3585
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.0061135919172755e-06,
+      "loss": 0.0005,
+      "step": 3590
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 5.95106288081197e-06,
+      "loss": 0.0003,
+      "step": 3595
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 5.89623151873957e-06,
+      "loss": 0.0002,
+      "step": 3600
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 5.841620137084691e-06,
+      "loss": 0.0408,
+      "step": 3605
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 5.787229364698904e-06,
+      "loss": 0.001,
+      "step": 3610
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 5.733059827893428e-06,
+      "loss": 0.0002,
+      "step": 3615
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 5.679112150431973e-06,
+      "loss": 0.0002,
+      "step": 3620
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 5.625386953523529e-06,
+      "loss": 0.0351,
+      "step": 3625
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 5.571884855815196e-06,
+      "loss": 0.0005,
+      "step": 3630
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 5.518606473385113e-06,
+      "loss": 0.0005,
+      "step": 3635
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 5.465552419735284e-06,
+      "loss": 0.0125,
+      "step": 3640
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 5.412723305784625e-06,
+      "loss": 0.0015,
+      "step": 3645
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 5.360119739861813e-06,
+      "loss": 0.0003,
+      "step": 3650
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 5.3077423276983566e-06,
+      "loss": 0.0005,
+      "step": 3655
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 5.2555916724215806e-06,
+      "loss": 0.0008,
+      "step": 3660
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 5.203668374547721e-06,
+      "loss": 0.0024,
+      "step": 3665
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 5.151973031974971e-06,
+      "loss": 0.0056,
+      "step": 3670
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 5.100506239976624e-06,
+      "loss": 0.0019,
+      "step": 3675
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 5.049268591194189e-06,
+      "loss": 0.0034,
+      "step": 3680
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 4.998260675630609e-06,
+      "loss": 0.014,
+      "step": 3685
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 4.947483080643411e-06,
+      "loss": 0.0001,
+      "step": 3690
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 4.8969363909380066e-06,
+      "loss": 0.0009,
+      "step": 3695
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 4.846621188560915e-06,
+      "loss": 0.0225,
+      "step": 3700
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 4.796538052893057e-06,
+      "loss": 0.0333,
+      "step": 3705
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 4.746687560643129e-06,
+      "loss": 0.01,
+      "step": 3710
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 4.69707028584089e-06,
+      "loss": 0.0012,
+      "step": 3715
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 4.64768679983065e-06,
+      "loss": 0.0004,
+      "step": 3720
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 4.59853767126458e-06,
+      "loss": 0.053,
+      "step": 3725
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 4.549623466096259e-06,
+      "loss": 0.0017,
+      "step": 3730
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 4.500944747574084e-06,
+      "loss": 0.0007,
+      "step": 3735
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 4.45250207623484e-06,
+      "loss": 0.0038,
+      "step": 3740
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.404296009897221e-06,
+      "loss": 0.0006,
+      "step": 3745
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.356327103655392e-06,
+      "loss": 0.0002,
+      "step": 3750
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.308595909872631e-06,
+      "loss": 0.0003,
+      "step": 3755
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 4.261102978174924e-06,
+      "loss": 0.0003,
+      "step": 3760
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 4.213848855444708e-06,
+      "loss": 0.0087,
+      "step": 3765
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.1668340858144835e-06,
+      "loss": 0.0002,
+      "step": 3770
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 4.120059210660618e-06,
+      "loss": 0.0009,
+      "step": 3775
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 4.073524768597075e-06,
+      "loss": 0.0021,
+      "step": 3780
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 4.027231295469233e-06,
+      "loss": 0.0011,
+      "step": 3785
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 3.981179324347705e-06,
+      "loss": 0.0018,
+      "step": 3790
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 3.935369385522197e-06,
+      "loss": 0.0003,
+      "step": 3795
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.889802006495416e-06,
+      "loss": 0.0021,
+      "step": 3800
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 3.844477711976971e-06,
+      "loss": 0.0097,
+      "step": 3805
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 3.79939702387736e-06,
+      "loss": 0.0086,
+      "step": 3810
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 3.7545604613019423e-06,
+      "loss": 0.0069,
+      "step": 3815
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 3.709968540544967e-06,
+      "loss": 0.0007,
+      "step": 3820
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 3.6656217750836166e-06,
+      "loss": 0.0007,
+      "step": 3825
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.6215206755721154e-06,
+      "loss": 0.0007,
+      "step": 3830
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.5776657498358346e-06,
+      "loss": 0.0004,
+      "step": 3835
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 3.5340575028654398e-06,
+      "loss": 0.0339,
+      "step": 3840
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 3.4906964368110972e-06,
+      "loss": 0.0001,
+      "step": 3845
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 3.4475830509766603e-06,
+      "loss": 0.0002,
+      "step": 3850
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.4047178418139563e-06,
+      "loss": 0.0045,
+      "step": 3855
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 3.3621013029170396e-06,
+      "loss": 0.0003,
+      "step": 3860
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 3.3197339250165297e-06,
+      "loss": 0.0002,
+      "step": 3865
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 3.2776161959739343e-06,
+      "loss": 0.0005,
+      "step": 3870
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 3.23574860077607e-06,
+      "loss": 0.013,
+      "step": 3875
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 3.1941316215294226e-06,
+      "loss": 0.0017,
+      "step": 3880
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 3.1527657374546714e-06,
+      "loss": 0.0003,
+      "step": 3885
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 3.1116514248810957e-06,
+      "loss": 0.0062,
+      "step": 3890
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 3.0707891572411247e-06,
+      "loss": 0.0369,
+      "step": 3895
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 3.030179405064901e-06,
+      "loss": 0.0005,
+      "step": 3900
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 2.989822635974815e-06,
+      "loss": 0.0102,
+      "step": 3905
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 2.949719314680194e-06,
+      "loss": 0.0078,
+      "step": 3910
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 2.909869902971865e-06,
+      "loss": 0.0036,
+      "step": 3915
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 2.8702748597169065e-06,
+      "loss": 0.0003,
+      "step": 3920
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 2.8309346408533084e-06,
+      "loss": 0.0233,
+      "step": 3925
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 2.791849699384791e-06,
+      "loss": 0.0009,
+      "step": 3930
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 2.75302048537551e-06,
+      "loss": 0.0007,
+      "step": 3935
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 2.7144474459449396e-06,
+      "loss": 0.0003,
+      "step": 3940
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 2.6761310252626767e-06,
+      "loss": 0.0011,
+      "step": 3945
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 2.6380716645433597e-06,
+      "loss": 0.0007,
+      "step": 3950
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 2.6002698020415743e-06,
+      "loss": 0.0022,
+      "step": 3955
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 2.562725873046795e-06,
+      "loss": 0.0005,
+      "step": 3960
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 2.525440309878399e-06,
+      "loss": 0.0021,
+      "step": 3965
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 2.488413541880663e-06,
+      "loss": 0.0005,
+      "step": 3970
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 2.4516459954178367e-06,
+      "loss": 0.0013,
+      "step": 3975
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 2.4151380938692265e-06,
+      "loss": 0.0019,
+      "step": 3980
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 2.378890257624322e-06,
+      "loss": 0.0001,
+      "step": 3985
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 2.3429029040779427e-06,
+      "loss": 0.0001,
+      "step": 3990
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 2.3071764476254613e-06,
+      "loss": 0.0019,
+      "step": 3995
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 2.2717112996579915e-06,
+      "loss": 0.0002,
+      "step": 4000
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 2.2365078685576885e-06,
+      "loss": 0.0007,
+      "step": 4005
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 2.2015665596930328e-06,
+      "loss": 0.001,
+      "step": 4010
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 2.166887775414142e-06,
+      "loss": 0.0003,
+      "step": 4015
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 2.132471915048173e-06,
+      "loss": 0.0009,
+      "step": 4020
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 2.0983193748947e-06,
+      "loss": 0.0009,
+      "step": 4025
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 2.064430548221166e-06,
+      "loss": 0.0009,
+      "step": 4030
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 2.030805825258325e-06,
+      "loss": 0.0117,
+      "step": 4035
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 1.9974455931957987e-06,
+      "loss": 0.0001,
+      "step": 4040
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 1.964350236177559e-06,
+      "loss": 0.0006,
+      "step": 4045
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 1.931520135297554e-06,
+      "loss": 0.0004,
+      "step": 4050
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 1.8989556685953008e-06,
+      "loss": 0.0005,
+      "step": 4055
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 1.8666572110515146e-06,
+      "loss": 0.0002,
+      "step": 4060
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 1.834625134583831e-06,
+      "loss": 0.0082,
+      "step": 4065
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 1.8028598080424758e-06,
+      "loss": 0.0369,
+      "step": 4070
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 1.7713615972060698e-06,
+      "loss": 0.0002,
+      "step": 4075
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 1.740130864777359e-06,
+      "loss": 0.0001,
+      "step": 4080
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 1.7091679703790942e-06,
+      "loss": 0.0007,
+      "step": 4085
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 1.6784732705498435e-06,
+      "loss": 0.0053,
+      "step": 4090
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 1.6480471187399233e-06,
+      "loss": 0.0001,
+      "step": 4095
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 1.6178898653073037e-06,
+      "loss": 0.0011,
+      "step": 4100
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 1.5880018575135825e-06,
+      "loss": 0.0007,
+      "step": 4105
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 1.558383439519992e-06,
+      "loss": 0.0001,
+      "step": 4110
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 1.529034952383418e-06,
+      "loss": 0.0001,
+      "step": 4115
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 1.4999567340525101e-06,
+      "loss": 0.0021,
+      "step": 4120
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 1.4711491193637322e-06,
+      "loss": 0.0251,
+      "step": 4125
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 1.4426124400375718e-06,
+      "loss": 0.0004,
+      "step": 4130
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 1.4143470246746588e-06,
+      "loss": 0.0144,
+      "step": 4135
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 1.3863531987520357e-06,
+      "loss": 0.0243,
+      "step": 4140
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 1.3586312846193716e-06,
+      "loss": 0.0001,
+      "step": 4145
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 1.3311816014952683e-06,
+      "loss": 0.006,
+      "step": 4150
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 1.3040044654635791e-06,
+      "loss": 0.0235,
+      "step": 4155
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 1.277100189469771e-06,
+      "loss": 0.0001,
+      "step": 4160
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 1.2504690833173156e-06,
+      "loss": 0.0005,
+      "step": 4165
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 1.2241114536641312e-06,
+      "loss": 0.0432,
+      "step": 4170
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 1.1980276040190502e-06,
+      "loss": 0.0002,
+      "step": 4175
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 1.1722178347383096e-06,
+      "loss": 0.0018,
+      "step": 4180
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 1.1466824430221123e-06,
+      "loss": 0.0003,
+      "step": 4185
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 1.1214217229111923e-06,
+      "loss": 0.0002,
+      "step": 4190
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 1.0964359652834404e-06,
+      "loss": 0.0065,
+      "step": 4195
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 1.0717254578505337e-06,
+      "loss": 0.0032,
+      "step": 4200
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 1.047290485154645e-06,
+      "loss": 0.0016,
+      "step": 4205
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 1.0231313285651522e-06,
+      "loss": 0.0023,
+      "step": 4210
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 9.992482662754044e-07,
+      "loss": 0.0064,
+      "step": 4215
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 9.756415732995267e-07,
+      "loss": 0.0034,
+      "step": 4220
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 9.523115214692236e-07,
+      "loss": 0.0004,
+      "step": 4225
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 9.29258379430692e-07,
+      "loss": 0.0001,
+      "step": 4230
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 9.064824126414823e-07,
+      "loss": 0.0366,
+      "step": 4235
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 8.839838833674896e-07,
+      "loss": 0.0002,
+      "step": 4240
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 8.617630506798868e-07,
+      "loss": 0.0001,
+      "step": 4245
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 8.398201704521741e-07,
+      "loss": 0.0002,
+      "step": 4250
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 8.181554953572146e-07,
+      "loss": 0.0005,
+      "step": 4255
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 7.967692748643396e-07,
+      "loss": 0.008,
+      "step": 4260
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 7.756617552364648e-07,
+      "loss": 0.0004,
+      "step": 4265
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 7.548331795272534e-07,
+      "loss": 0.001,
+      "step": 4270
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 7.342837875783298e-07,
+      "loss": 0.042,
+      "step": 4275
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 7.140138160164983e-07,
+      "loss": 0.0012,
+      "step": 4280
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 6.940234982510452e-07,
+      "loss": 0.0279,
+      "step": 4285
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 6.743130644710216e-07,
+      "loss": 0.0026,
+      "step": 4290
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 6.548827416426151e-07,
+      "loss": 0.0007,
+      "step": 4295
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 6.357327535065238e-07,
+      "loss": 0.0004,
+      "step": 4300
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 6.168633205753893e-07,
+      "loss": 0.0095,
+      "step": 4305
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 5.982746601312456e-07,
+      "loss": 0.024,
+      "step": 4310
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 5.799669862230267e-07,
+      "loss": 0.0003,
+      "step": 4315
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 5.619405096641023e-07,
+      "loss": 0.0141,
+      "step": 4320
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 5.441954380298403e-07,
+      "loss": 0.0004,
+      "step": 4325
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 5.267319756552286e-07,
+      "loss": 0.0008,
+      "step": 4330
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 5.095503236325156e-07,
+      "loss": 0.0015,
+      "step": 4335
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 4.926506798088954e-07,
+      "loss": 0.0004,
+      "step": 4340
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 4.760332387842292e-07,
+      "loss": 0.0153,
+      "step": 4345
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 4.596981919088056e-07,
+      "loss": 0.0105,
+      "step": 4350
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 4.436457272811334e-07,
+      "loss": 0.0032,
+      "step": 4355
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 4.2787602974578846e-07,
+      "loss": 0.0001,
+      "step": 4360
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 4.123892808912677e-07,
+      "loss": 0.0197,
+      "step": 4365
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 3.971856590479078e-07,
+      "loss": 0.0003,
+      "step": 4370
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 3.822653392858311e-07,
+      "loss": 0.0,
+      "step": 4375
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 3.6762849341293036e-07,
+      "loss": 0.0026,
+      "step": 4380
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 3.5327528997288474e-07,
+      "loss": 0.0232,
+      "step": 4385
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 3.392058942432247e-07,
+      "loss": 0.0002,
+      "step": 4390
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 3.2542046823343107e-07,
+      "loss": 0.0002,
+      "step": 4395
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 3.11919170683056e-07,
+      "loss": 0.034,
+      "step": 4400
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 2.9870215705991576e-07,
+      "loss": 0.0007,
+      "step": 4405
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 2.857695795582843e-07,
+      "loss": 0.0009,
+      "step": 4410
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 2.731215870971443e-07,
+      "loss": 0.0031,
+      "step": 4415
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 2.6075832531847753e-07,
+      "loss": 0.0002,
+      "step": 4420
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 2.4867993658557744e-07,
+      "loss": 0.0139,
+      "step": 4425
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 2.3688655998142516e-07,
+      "loss": 0.0015,
+      "step": 4430
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 2.2537833130707454e-07,
+      "loss": 0.0369,
+      "step": 4435
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 2.141553830800974e-07,
+      "loss": 0.0001,
+      "step": 4440
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 2.032178445330435e-07,
+      "loss": 0.0001,
+      "step": 4445
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 1.925658416119719e-07,
+      "loss": 0.0032,
+      "step": 4450
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 1.821994969749885e-07,
+      "loss": 0.0002,
+      "step": 4455
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 1.7211892999083877e-07,
+      "loss": 0.0001,
+      "step": 4460
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 1.6232425673752272e-07,
+      "loss": 0.0001,
+      "step": 4465
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 1.5281559000097367e-07,
+      "loss": 0.0002,
+      "step": 4470
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 1.4359303927374833e-07,
+      "loss": 0.0254,
+      "step": 4475
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 1.3465671075377218e-07,
+      "loss": 0.0004,
+      "step": 4480
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 1.2600670734311536e-07,
+      "loss": 0.0006,
+      "step": 4485
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 1.1764312864679938e-07,
+      "loss": 0.0365,
+      "step": 4490
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 1.095660709716645e-07,
+      "loss": 0.0,
+      "step": 4495
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 1.0177562732525125e-07,
+      "loss": 0.0002,
+      "step": 4500
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 9.427188741473192e-08,
+      "loss": 0.0004,
+      "step": 4505
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 8.705493764587513e-08,
+      "loss": 0.0002,
+      "step": 4510
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 8.012486112205785e-08,
+      "loss": 0.0003,
+      "step": 4515
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 7.348173764329947e-08,
+      "loss": 0.0189,
+      "step": 4520
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 6.712564370535135e-08,
+      "loss": 0.0005,
+      "step": 4525
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 6.105665249880599e-08,
+      "loss": 0.0271,
+      "step": 4530
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 5.5274833908267044e-08,
+      "loss": 0.0001,
+      "step": 4535
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 4.9780254511530563e-08,
+      "loss": 0.0051,
+      "step": 4540
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 4.457297757882728e-08,
+      "loss": 0.0001,
+      "step": 4545
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 3.965306307210093e-08,
+      "loss": 0.0001,
+      "step": 4550
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 3.5020567644294954e-08,
+      "loss": 0.0013,
+      "step": 4555
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 3.067554463872524e-08,
+      "loss": 0.0001,
+      "step": 4560
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 2.6618044088450034e-08,
+      "loss": 0.0077,
+      "step": 4565
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 2.28481127156982e-08,
+      "loss": 0.0006,
+      "step": 4570
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 1.9365793931330754e-08,
+      "loss": 0.0001,
+      "step": 4575
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 1.61711278343496e-08,
+      "loss": 0.0003,
+      "step": 4580
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 1.326415121141733e-08,
+      "loss": 0.0003,
+      "step": 4585
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 1.0644897536452037e-08,
+      "loss": 0.0069,
+      "step": 4590
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 8.313396970233145e-09,
+      "loss": 0.0172,
+      "step": 4595
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 6.269676360046162e-09,
+      "loss": 0.0002,
+      "step": 4600
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 4.513759239399562e-09,
+      "loss": 0.0001,
+      "step": 4605
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 3.0456658277194793e-09,
+      "loss": 0.002,
+      "step": 4610
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 1.8654130301470894e-09,
+      "loss": 0.0128,
+      "step": 4615
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 9.730144373221172e-10,
+      "loss": 0.0041,
+      "step": 4620
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 3.6848032524128275e-10,
+      "loss": 0.0182,
+      "step": 4625
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 5.181765512785042e-11,
+      "loss": 0.0123,
+      "step": 4630
+    },
+    {
+      "epoch": 6.0,
+      "step": 4632,
+      "total_flos": 6.270474759151288e+17,
+      "train_loss": 0.07220351306288228,
+      "train_runtime": 14592.326,
+      "train_samples_per_second": 5.081,
+      "train_steps_per_second": 0.317
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 4632,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 6,
+  "save_steps": 1000,
+  "total_flos": 6.270474759151288e+17,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}