diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,31439 @@
+{
+  "best_metric": 23.848435348153973,
+  "best_model_checkpoint": "/content/XLMFinetune/model-bin2/test/checkpoint-26168",
+  "epoch": 2.0,
+  "eval_steps": 500,
+  "global_step": 26168,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.111535523300229e-07,
+      "loss": 5.3672,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3750954927425516e-06,
+      "loss": 5.4479,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1390374331550802e-06,
+      "loss": 5.3383,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.902979373567609e-06,
+      "loss": 4.952,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.666921313980138e-06,
+      "loss": 4.4477,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.430863254392666e-06,
+      "loss": 4.4078,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.194805194805195e-06,
+      "loss": 4.5465,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.958747135217724e-06,
+      "loss": 4.3111,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.722689075630252e-06,
+      "loss": 4.3164,
+      "step": 45
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.4866310160427806e-06,
+      "loss": 4.1561,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.25057295645531e-06,
+      "loss": 5.8121,
+      "step": 55
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.014514896867839e-06,
+      "loss": 5.9082,
+      "step": 60
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.778456837280367e-06,
+      "loss": 5.732,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0542398777692896e-05,
+      "loss": 5.7656,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1306340718105425e-05,
+      "loss": 5.3199,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2070282658517953e-05,
+      "loss": 5.1668,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2834224598930484e-05,
+      "loss": 5.2023,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.359816653934301e-05,
+      "loss": 4.1394,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.436210847975554e-05,
+      "loss": 3.974,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5126050420168067e-05,
+      "loss": 3.4523,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5889992360580598e-05,
+      "loss": 5.4578,
+      "step": 105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6653934300993127e-05,
+      "loss": 5.7855,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7417876241405653e-05,
+      "loss": 5.0931,
+      "step": 115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8181818181818182e-05,
+      "loss": 5.3789,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.894576012223071e-05,
+      "loss": 5.2285,
+      "step": 125
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.970970206264324e-05,
+      "loss": 4.8857,
+      "step": 130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.047364400305577e-05,
+      "loss": 4.5129,
+      "step": 135
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1237585943468296e-05,
+      "loss": 4.0876,
+      "step": 140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2001527883880825e-05,
+      "loss": 4.2938,
+      "step": 145
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2765469824293358e-05,
+      "loss": 3.9873,
+      "step": 150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3529411764705884e-05,
+      "loss": 5.5092,
+      "step": 155
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4293353705118413e-05,
+      "loss": 5.9281,
+      "step": 160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5057295645530942e-05,
+      "loss": 5.6781,
+      "step": 165
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5821237585943468e-05,
+      "loss": 4.7973,
+      "step": 170
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6585179526355997e-05,
+      "loss": 5.1039,
+      "step": 175
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.734912146676853e-05,
+      "loss": 5.259,
+      "step": 180
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8113063407181056e-05,
+      "loss": 5.0793,
+      "step": 185
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8877005347593582e-05,
+      "loss": 4.8102,
+      "step": 190
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9640947288006115e-05,
+      "loss": 4.2803,
+      "step": 195
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0404889228418644e-05,
+      "loss": 3.758,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1168831168831166e-05,
+      "loss": 5.957,
+      "step": 205
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1932773109243696e-05,
+      "loss": 5.9391,
+      "step": 210
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.269671504965623e-05,
+      "loss": 5.4367,
+      "step": 215
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3460656990068754e-05,
+      "loss": 5.3895,
+      "step": 220
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4224598930481284e-05,
+      "loss": 5.2668,
+      "step": 225
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.498854087089381e-05,
+      "loss": 4.9992,
+      "step": 230
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.575248281130634e-05,
+      "loss": 4.3316,
+      "step": 235
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.651642475171887e-05,
+      "loss": 4.4064,
+      "step": 240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.72803666921314e-05,
+      "loss": 3.9166,
+      "step": 245
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.804430863254393e-05,
+      "loss": 3.5941,
+      "step": 250
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.880825057295646e-05,
+      "loss": 5.6594,
+      "step": 255
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.957219251336899e-05,
+      "loss": 5.5082,
+      "step": 260
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.033613445378152e-05,
+      "loss": 5.4387,
+      "step": 265
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.110007639419404e-05,
+      "loss": 5.1539,
+      "step": 270
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.186401833460657e-05,
+      "loss": 5.0574,
+      "step": 275
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.26279602750191e-05,
+      "loss": 4.8289,
+      "step": 280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.339190221543163e-05,
+      "loss": 4.8527,
+      "step": 285
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.415584415584416e-05,
+      "loss": 4.3766,
+      "step": 290
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.491978609625669e-05,
+      "loss": 3.7943,
+      "step": 295
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5683728036669216e-05,
+      "loss": 3.2498,
+      "step": 300
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6447669977081745e-05,
+      "loss": 5.7168,
+      "step": 305
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.7211611917494275e-05,
+      "loss": 5.8766,
+      "step": 310
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.7975553857906804e-05,
+      "loss": 5.7383,
+      "step": 315
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8739495798319326e-05,
+      "loss": 5.4395,
+      "step": 320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.950343773873186e-05,
+      "loss": 5.3113,
+      "step": 325
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.026737967914439e-05,
+      "loss": 4.9613,
+      "step": 330
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.1031321619556914e-05,
+      "loss": 4.5619,
+      "step": 335
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.1795263559969444e-05,
+      "loss": 4.3414,
+      "step": 340
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.255920550038197e-05,
+      "loss": 4.4539,
+      "step": 345
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.332314744079451e-05,
+      "loss": 3.3832,
+      "step": 350
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.408708938120703e-05,
+      "loss": 5.9297,
+      "step": 355
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.485103132161956e-05,
+      "loss": 5.4016,
+      "step": 360
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.561497326203209e-05,
+      "loss": 4.8334,
+      "step": 365
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.637891520244461e-05,
+      "loss": 5.2273,
+      "step": 370
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.714285714285714e-05,
+      "loss": 4.9238,
+      "step": 375
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.7754010695187164e-05,
+      "loss": 4.3672,
+      "step": 380
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.851795263559969e-05,
+      "loss": 4.2797,
+      "step": 385
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.928189457601223e-05,
+      "loss": 4.1217,
+      "step": 390
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.004583651642476e-05,
+      "loss": 4.0648,
+      "step": 395
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.080977845683729e-05,
+      "loss": 3.3252,
+      "step": 400
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.157372039724982e-05,
+      "loss": 5.5906,
+      "step": 405
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.233766233766233e-05,
+      "loss": 5.3297,
+      "step": 410
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.310160427807486e-05,
+      "loss": 5.5746,
+      "step": 415
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.386554621848739e-05,
+      "loss": 5.2078,
+      "step": 420
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.462948815889993e-05,
+      "loss": 4.8525,
+      "step": 425
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.539343009931246e-05,
+      "loss": 4.7914,
+      "step": 430
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.615737203972498e-05,
+      "loss": 4.6715,
+      "step": 435
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.692131398013751e-05,
+      "loss": 4.5393,
+      "step": 440
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.753246753246754e-05,
+      "loss": 4.075,
+      "step": 445
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.829640947288007e-05,
+      "loss": 3.6494,
+      "step": 450
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.90603514132926e-05,
+      "loss": 5.5176,
+      "step": 455
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.967150496562261e-05,
+      "loss": 6.3191,
+      "step": 460
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.043544690603514e-05,
+      "loss": 4.6111,
+      "step": 465
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.119938884644768e-05,
+      "loss": 5.2383,
+      "step": 470
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.181054239877769e-05,
+      "loss": 5.5373,
+      "step": 475
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.257448433919023e-05,
+      "loss": 4.9678,
+      "step": 480
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.333842627960276e-05,
+      "loss": 4.3086,
+      "step": 485
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.410236822001529e-05,
+      "loss": 4.2645,
+      "step": 490
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.486631016042782e-05,
+      "loss": 3.6471,
+      "step": 495
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.563025210084033e-05,
+      "loss": 2.734,
+      "step": 500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.639419404125286e-05,
+      "loss": 5.6322,
+      "step": 505
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.71581359816654e-05,
+      "loss": 6.4516,
+      "step": 510
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.792207792207793e-05,
+      "loss": 6.141,
+      "step": 515
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.868601986249046e-05,
+      "loss": 4.4348,
+      "step": 520
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.944996180290298e-05,
+      "loss": 4.6922,
+      "step": 525
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.021390374331551e-05,
+      "loss": 4.8404,
+      "step": 530
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.097784568372804e-05,
+      "loss": 5.326,
+      "step": 535
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.158899923605807e-05,
+      "loss": 4.1762,
+      "step": 540
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.23529411764706e-05,
+      "loss": 2.7068,
+      "step": 545
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.311688311688312e-05,
+      "loss": 2.9233,
+      "step": 550
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.388082505729565e-05,
+      "loss": 5.177,
+      "step": 555
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.464476699770818e-05,
+      "loss": 5.1857,
+      "step": 560
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.540870893812071e-05,
+      "loss": 4.8047,
+      "step": 565
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.617265087853324e-05,
+      "loss": 5.1225,
+      "step": 570
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.693659281894576e-05,
+      "loss": 3.5857,
+      "step": 575
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.770053475935829e-05,
+      "loss": 5.016,
+      "step": 580
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.846447669977083e-05,
+      "loss": 4.1475,
+      "step": 585
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.922841864018336e-05,
+      "loss": 3.4462,
+      "step": 590
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.999236058059587e-05,
+      "loss": 2.926,
+      "step": 595
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.07563025210084e-05,
+      "loss": 3.36,
+      "step": 600
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.152024446142093e-05,
+      "loss": 5.6203,
+      "step": 605
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.228418640183346e-05,
+      "loss": 5.2746,
+      "step": 610
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.3048128342246e-05,
+      "loss": 6.3246,
+      "step": 615
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.381207028265852e-05,
+      "loss": 4.7615,
+      "step": 620
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.457601222307105e-05,
+      "loss": 4.4709,
+      "step": 625
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.533995416348358e-05,
+      "loss": 4.0367,
+      "step": 630
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.610389610389611e-05,
+      "loss": 4.5443,
+      "step": 635
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.686783804430864e-05,
+      "loss": 2.754,
+      "step": 640
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.763177998472117e-05,
+      "loss": 2.841,
+      "step": 645
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.83957219251337e-05,
+      "loss": 2.6588,
+      "step": 650
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.915966386554623e-05,
+      "loss": 5.3617,
+      "step": 655
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992360580595875e-05,
+      "loss": 6.0061,
+      "step": 660
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00010068754774637128,
+      "loss": 5.8316,
+      "step": 665
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00010145148968678381,
+      "loss": 5.1043,
+      "step": 670
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00010221543162719633,
+      "loss": 3.8354,
+      "step": 675
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00010297937356760887,
+      "loss": 3.8439,
+      "step": 680
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00010374331550802139,
+      "loss": 5.7764,
+      "step": 685
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00010450725744843393,
+      "loss": 3.8581,
+      "step": 690
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00010527119938884645,
+      "loss": 2.841,
+      "step": 695
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00010603514132925898,
+      "loss": 2.4475,
+      "step": 700
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00010679908326967152,
+      "loss": 5.5125,
+      "step": 705
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00010756302521008403,
+      "loss": 5.4578,
+      "step": 710
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00010832696715049658,
+      "loss": 5.5195,
+      "step": 715
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00010909090909090909,
+      "loss": 4.5291,
+      "step": 720
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00010985485103132162,
+      "loss": 4.5887,
+      "step": 725
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00011061879297173416,
+      "loss": 5.2873,
+      "step": 730
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00011138273491214668,
+      "loss": 3.6059,
+      "step": 735
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00011214667685255922,
+      "loss": 3.4224,
+      "step": 740
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00011291061879297174,
+      "loss": 3.1507,
+      "step": 745
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00011367456073338427,
+      "loss": 2.7281,
+      "step": 750
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00011443850267379678,
+      "loss": 4.7617,
+      "step": 755
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00011520244461420933,
+      "loss": 5.1021,
+      "step": 760
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00011596638655462187,
+      "loss": 4.8506,
+      "step": 765
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00011673032849503439,
+      "loss": 4.4602,
+      "step": 770
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00011749427043544691,
+      "loss": 3.5974,
+      "step": 775
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00011825821237585943,
+      "loss": 3.5879,
+      "step": 780
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00011902215431627197,
+      "loss": 4.2986,
+      "step": 785
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00011978609625668449,
+      "loss": 2.5129,
+      "step": 790
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00012055003819709703,
+      "loss": 3.3521,
+      "step": 795
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00012131398013750956,
+      "loss": 1.9732,
+      "step": 800
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00012207792207792208,
+      "loss": 4.8342,
+      "step": 805
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001228418640183346,
+      "loss": 5.3076,
+      "step": 810
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00012360580595874714,
+      "loss": 4.8811,
+      "step": 815
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00012436974789915966,
+      "loss": 5.3889,
+      "step": 820
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001251336898395722,
+      "loss": 3.7623,
+      "step": 825
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00012589763177998472,
+      "loss": 4.561,
+      "step": 830
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00012666157372039725,
+      "loss": 4.0161,
+      "step": 835
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00012742551566080978,
+      "loss": 3.5828,
+      "step": 840
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001281894576012223,
+      "loss": 2.6784,
+      "step": 845
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00012895339954163484,
+      "loss": 2.1759,
+      "step": 850
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00012971734148204737,
+      "loss": 5.285,
+      "step": 855
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001304812834224599,
+      "loss": 4.3326,
+      "step": 860
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00013124522536287243,
+      "loss": 4.8477,
+      "step": 865
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00013200916730328496,
+      "loss": 3.8988,
+      "step": 870
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001327731092436975,
+      "loss": 3.8404,
+      "step": 875
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00013353705118411002,
+      "loss": 2.9827,
+      "step": 880
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00013430099312452255,
+      "loss": 3.5191,
+      "step": 885
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00013506493506493507,
+      "loss": 3.0866,
+      "step": 890
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001358288770053476,
+      "loss": 3.6331,
+      "step": 895
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00013659281894576013,
+      "loss": 2.1087,
+      "step": 900
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00013735676088617266,
+      "loss": 4.9768,
+      "step": 905
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001381207028265852,
+      "loss": 4.974,
+      "step": 910
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00013888464476699772,
+      "loss": 5.1777,
+      "step": 915
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00013964858670741022,
+      "loss": 5.3486,
+      "step": 920
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00014041252864782278,
+      "loss": 4.2877,
+      "step": 925
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001411764705882353,
+      "loss": 3.8871,
+      "step": 930
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00014194041252864784,
+      "loss": 3.4246,
+      "step": 935
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00014270435446906037,
+      "loss": 4.7645,
+      "step": 940
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00014346829640947287,
+      "loss": 2.9158,
+      "step": 945
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00014423223834988543,
+      "loss": 1.6789,
+      "step": 950
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00014499618029029793,
+      "loss": 4.8162,
+      "step": 955
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00014576012223071048,
+      "loss": 4.7348,
+      "step": 960
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00014652406417112301,
+      "loss": 4.8879,
+      "step": 965
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00014728800611153552,
+      "loss": 6.2023,
+      "step": 970
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00014805194805194807,
+      "loss": 3.7848,
+      "step": 975
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00014881588999236057,
+      "loss": 3.9518,
+      "step": 980
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00014957983193277313,
+      "loss": 5.0244,
+      "step": 985
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00015034377387318566,
+      "loss": 3.1289,
+      "step": 990
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00015110771581359816,
+      "loss": 2.6522,
+      "step": 995
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00015187165775401072,
+      "loss": 2.3406,
+      "step": 1000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00015263559969442322,
+      "loss": 5.4945,
+      "step": 1005
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00015339954163483578,
+      "loss": 6.2398,
+      "step": 1010
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00015416348357524828,
+      "loss": 4.6941,
+      "step": 1015
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001549274255156608,
+      "loss": 5.5852,
+      "step": 1020
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00015569136745607337,
+      "loss": 4.1475,
+      "step": 1025
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00015645530939648587,
+      "loss": 3.8572,
+      "step": 1030
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001572192513368984,
+      "loss": 3.3049,
+      "step": 1035
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00015798319327731093,
+      "loss": 2.8052,
+      "step": 1040
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00015874713521772346,
+      "loss": 3.4445,
+      "step": 1045
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00015951107715813598,
+      "loss": 1.85,
+      "step": 1050
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016027501909854851,
+      "loss": 5.0895,
+      "step": 1055
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016103896103896104,
+      "loss": 4.1975,
+      "step": 1060
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016180290297937357,
+      "loss": 4.8666,
+      "step": 1065
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001625668449197861,
+      "loss": 5.3641,
+      "step": 1070
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016333078686019863,
+      "loss": 4.3744,
+      "step": 1075
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016409472880061116,
+      "loss": 4.8926,
+      "step": 1080
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001648586707410237,
+      "loss": 3.8107,
+      "step": 1085
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016562261268143622,
+      "loss": 2.8379,
+      "step": 1090
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016638655462184875,
+      "loss": 2.4593,
+      "step": 1095
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016715049656226128,
+      "loss": 2.2865,
+      "step": 1100
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001679144385026738,
+      "loss": 5.5463,
+      "step": 1105
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016867838044308634,
+      "loss": 5.9701,
+      "step": 1110
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00016944232238349887,
+      "loss": 4.5203,
+      "step": 1115
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001702062643239114,
+      "loss": 4.7477,
+      "step": 1120
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00017097020626432392,
+      "loss": 3.3791,
+      "step": 1125
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00017173414820473645,
+      "loss": 3.7066,
+      "step": 1130
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00017249809014514898,
+      "loss": 4.1764,
+      "step": 1135
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001732620320855615,
+      "loss": 3.2178,
+      "step": 1140
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00017402597402597401,
+      "loss": 2.541,
+      "step": 1145
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00017478991596638657,
+      "loss": 3.3618,
+      "step": 1150
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001755538579067991,
+      "loss": 4.0588,
+      "step": 1155
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00017631779984721163,
+      "loss": 5.4387,
+      "step": 1160
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00017708174178762416,
+      "loss": 4.2572,
+      "step": 1165
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00017784568372803666,
+      "loss": 5.133,
+      "step": 1170
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00017860962566844922,
+      "loss": 4.4044,
+      "step": 1175
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00017937356760886172,
+      "loss": 3.218,
+      "step": 1180
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00018013750954927428,
+      "loss": 3.2953,
+      "step": 1185
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001809014514896868,
+      "loss": 2.4731,
+      "step": 1190
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001816653934300993,
+      "loss": 3.5734,
+      "step": 1195
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00018242933537051186,
+      "loss": 1.5107,
+      "step": 1200
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00018319327731092437,
+      "loss": 4.7984,
+      "step": 1205
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00018395721925133692,
+      "loss": 5.5371,
+      "step": 1210
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00018472116119174942,
+      "loss": 3.9248,
+      "step": 1215
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00018548510313216195,
+      "loss": 4.2715,
+      "step": 1220
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001862490450725745,
+      "loss": 4.9904,
+      "step": 1225
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000187012987012987,
+      "loss": 3.2785,
+      "step": 1230
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00018777692895339957,
+      "loss": 3.5854,
+      "step": 1235
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00018854087089381207,
+      "loss": 3.5434,
+      "step": 1240
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001893048128342246,
+      "loss": 2.4767,
+      "step": 1245
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019006875477463716,
+      "loss": 1.1776,
+      "step": 1250
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019083269671504966,
+      "loss": 5.5906,
+      "step": 1255
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019159663865546221,
+      "loss": 5.1738,
+      "step": 1260
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019236058059587472,
+      "loss": 4.7447,
+      "step": 1265
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019312452253628725,
+      "loss": 4.5672,
+      "step": 1270
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019388846447669978,
+      "loss": 3.4076,
+      "step": 1275
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001946524064171123,
+      "loss": 3.9701,
+      "step": 1280
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019541634835752486,
+      "loss": 3.6115,
+      "step": 1285
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019618029029793736,
+      "loss": 3.1557,
+      "step": 1290
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001969442322383499,
+      "loss": 2.698,
+      "step": 1295
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019770817417876242,
+      "loss": 1.5998,
+      "step": 1300
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019847211611917495,
+      "loss": 5.1922,
+      "step": 1305
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019923605805958748,
+      "loss": 4.9844,
+      "step": 1310
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 3.5535,
+      "step": 1315
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019995977312039907,
+      "loss": 3.9018,
+      "step": 1320
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019991954624079813,
+      "loss": 4.1891,
+      "step": 1325
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019987931936119716,
+      "loss": 3.2686,
+      "step": 1330
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001998390924815962,
+      "loss": 3.498,
+      "step": 1335
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019979886560199525,
+      "loss": 1.9607,
+      "step": 1340
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001997586387223943,
+      "loss": 2.2703,
+      "step": 1345
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019971841184279336,
+      "loss": 2.7828,
+      "step": 1350
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019967818496319242,
+      "loss": 5.5039,
+      "step": 1355
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019963795808359148,
+      "loss": 5.6221,
+      "step": 1360
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001995977312039905,
+      "loss": 5.0553,
+      "step": 1365
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019955750432438957,
+      "loss": 4.3732,
+      "step": 1370
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001995172774447886,
+      "loss": 4.2107,
+      "step": 1375
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019947705056518766,
+      "loss": 3.9906,
+      "step": 1380
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019943682368558672,
+      "loss": 3.2902,
+      "step": 1385
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019939659680598577,
+      "loss": 3.6535,
+      "step": 1390
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019935636992638483,
+      "loss": 3.9422,
+      "step": 1395
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019931614304678386,
+      "loss": 1.6013,
+      "step": 1400
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019927591616718292,
+      "loss": 5.3186,
+      "step": 1405
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019923568928758195,
+      "loss": 4.8236,
+      "step": 1410
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000199195462407981,
+      "loss": 4.2941,
+      "step": 1415
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019915523552838007,
+      "loss": 3.9504,
+      "step": 1420
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019911500864877913,
+      "loss": 3.9039,
+      "step": 1425
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001990747817691782,
+      "loss": 3.1057,
+      "step": 1430
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019903455488957722,
+      "loss": 3.3202,
+      "step": 1435
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019899432800997628,
+      "loss": 4.9787,
+      "step": 1440
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019895410113037533,
+      "loss": 2.4636,
+      "step": 1445
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019891387425077437,
+      "loss": 2.2541,
+      "step": 1450
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019887364737117342,
+      "loss": 4.0754,
+      "step": 1455
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019883342049157248,
+      "loss": 5.0287,
+      "step": 1460
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019879319361197154,
+      "loss": 4.4176,
+      "step": 1465
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019875296673237057,
+      "loss": 4.0273,
+      "step": 1470
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019871273985276963,
+      "loss": 4.6488,
+      "step": 1475
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001986725129731687,
+      "loss": 3.7554,
+      "step": 1480
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019863228609356772,
+      "loss": 4.065,
+      "step": 1485
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019859205921396678,
+      "loss": 3.813,
+      "step": 1490
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019855183233436584,
+      "loss": 2.2622,
+      "step": 1495
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001985116054547649,
+      "loss": 1.2104,
+      "step": 1500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019847137857516392,
+      "loss": 4.7545,
+      "step": 1505
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019843115169556298,
+      "loss": 5.3623,
+      "step": 1510
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019839092481596204,
+      "loss": 3.8675,
+      "step": 1515
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001983506979363611,
+      "loss": 3.4828,
+      "step": 1520
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019831047105676013,
+      "loss": 3.7339,
+      "step": 1525
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001982702441771592,
+      "loss": 3.509,
+      "step": 1530
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019823001729755825,
+      "loss": 3.3487,
+      "step": 1535
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019818979041795728,
+      "loss": 5.2811,
+      "step": 1540
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019814956353835634,
+      "loss": 2.0317,
+      "step": 1545
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001981093366587554,
+      "loss": 1.7232,
+      "step": 1550
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019806910977915445,
+      "loss": 3.9217,
+      "step": 1555
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019802888289955348,
+      "loss": 4.3631,
+      "step": 1560
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019798865601995254,
+      "loss": 6.2506,
+      "step": 1565
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019795647451627176,
+      "loss": 4.4891,
+      "step": 1570
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019791624763667082,
+      "loss": 3.79,
+      "step": 1575
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019787602075706988,
+      "loss": 4.5018,
+      "step": 1580
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019783579387746894,
+      "loss": 3.6576,
+      "step": 1585
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000197795566997868,
+      "loss": 3.0058,
+      "step": 1590
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019775534011826705,
+      "loss": 3.6505,
+      "step": 1595
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019771511323866608,
+      "loss": 1.5467,
+      "step": 1600
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019767488635906514,
+      "loss": 4.4375,
+      "step": 1605
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019763465947946417,
+      "loss": 4.4432,
+      "step": 1610
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019759443259986323,
+      "loss": 4.8396,
+      "step": 1615
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001975542057202623,
+      "loss": 4.9068,
+      "step": 1620
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019751397884066135,
+      "loss": 3.3321,
+      "step": 1625
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001974737519610604,
+      "loss": 3.0243,
+      "step": 1630
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019743352508145944,
+      "loss": 3.6723,
+      "step": 1635
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001973932982018585,
+      "loss": 3.564,
+      "step": 1640
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019735307132225753,
+      "loss": 1.2411,
+      "step": 1645
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019731284444265659,
+      "loss": 0.7007,
+      "step": 1650
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019727261756305564,
+      "loss": 4.9119,
+      "step": 1655
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001972323906834547,
+      "loss": 6.3875,
+      "step": 1660
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019719216380385376,
+      "loss": 4.9604,
+      "step": 1665
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001971519369242528,
+      "loss": 5.1688,
+      "step": 1670
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019711171004465185,
+      "loss": 3.3456,
+      "step": 1675
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001970714831650509,
+      "loss": 4.251,
+      "step": 1680
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019703125628544994,
+      "loss": 2.4675,
+      "step": 1685
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000196991029405849,
+      "loss": 3.2748,
+      "step": 1690
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019695080252624806,
+      "loss": 2.7843,
+      "step": 1695
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019691057564664709,
+      "loss": 2.922,
+      "step": 1700
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019687034876704614,
+      "loss": 4.8305,
+      "step": 1705
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001968301218874452,
+      "loss": 4.2068,
+      "step": 1710
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019678989500784426,
+      "loss": 4.2574,
+      "step": 1715
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001967496681282433,
+      "loss": 3.9824,
+      "step": 1720
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019670944124864235,
+      "loss": 4.5187,
+      "step": 1725
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001966692143690414,
+      "loss": 3.1695,
+      "step": 1730
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019662898748944044,
+      "loss": 2.4495,
+      "step": 1735
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001965887606098395,
+      "loss": 3.9741,
+      "step": 1740
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019654853373023856,
+      "loss": 3.6335,
+      "step": 1745
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019650830685063761,
+      "loss": 1.5263,
+      "step": 1750
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019646807997103667,
+      "loss": 4.2363,
+      "step": 1755
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001964278530914357,
+      "loss": 4.9076,
+      "step": 1760
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019638762621183476,
+      "loss": 4.7342,
+      "step": 1765
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001963473993322338,
+      "loss": 6.3031,
+      "step": 1770
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019630717245263285,
+      "loss": 4.618,
+      "step": 1775
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001962669455730319,
+      "loss": 3.8465,
+      "step": 1780
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019622671869343097,
+      "loss": 2.1621,
+      "step": 1785
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019618649181383003,
+      "loss": 2.1442,
+      "step": 1790
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019614626493422906,
+      "loss": 2.3709,
+      "step": 1795
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019610603805462812,
+      "loss": 3.6523,
+      "step": 1800
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019606581117502715,
+      "loss": 5.8377,
+      "step": 1805
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001960255842954262,
+      "loss": 4.6049,
+      "step": 1810
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019598535741582526,
+      "loss": 4.4676,
+      "step": 1815
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019594513053622432,
+      "loss": 4.2148,
+      "step": 1820
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019590490365662338,
+      "loss": 4.3605,
+      "step": 1825
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019586467677702244,
+      "loss": 3.4217,
+      "step": 1830
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019582444989742147,
+      "loss": 4.5377,
+      "step": 1835
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001957842230178205,
+      "loss": 2.128,
+      "step": 1840
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019574399613821956,
+      "loss": 2.0646,
+      "step": 1845
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019570376925861862,
+      "loss": 2.895,
+      "step": 1850
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019566354237901768,
+      "loss": 4.5703,
+      "step": 1855
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019562331549941673,
+      "loss": 4.5592,
+      "step": 1860
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001955830886198158,
+      "loss": 5.3285,
+      "step": 1865
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019554286174021482,
+      "loss": 4.4393,
+      "step": 1870
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019550263486061385,
+      "loss": 4.2725,
+      "step": 1875
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001954624079810129,
+      "loss": 2.5464,
+      "step": 1880
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019542218110141197,
+      "loss": 3.3585,
+      "step": 1885
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019538195422181103,
+      "loss": 2.9919,
+      "step": 1890
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001953417273422101,
+      "loss": 1.8249,
+      "step": 1895
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019530150046260915,
+      "loss": 2.5888,
+      "step": 1900
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019526127358300818,
+      "loss": 5.8953,
+      "step": 1905
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001952210467034072,
+      "loss": 5.2256,
+      "step": 1910
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019518081982380627,
+      "loss": 4.6908,
+      "step": 1915
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019514059294420532,
+      "loss": 3.9816,
+      "step": 1920
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019510036606460438,
+      "loss": 3.4984,
+      "step": 1925
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019506013918500344,
+      "loss": 2.786,
+      "step": 1930
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001950199123054025,
+      "loss": 2.0924,
+      "step": 1935
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019497968542580153,
+      "loss": 2.1778,
+      "step": 1940
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019493945854620056,
+      "loss": 3.1165,
+      "step": 1945
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019489923166659962,
+      "loss": 1.3559,
+      "step": 1950
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019485900478699868,
+      "loss": 4.5992,
+      "step": 1955
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019481877790739774,
+      "loss": 4.9736,
+      "step": 1960
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001947785510277968,
+      "loss": 5.0469,
+      "step": 1965
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019473832414819585,
+      "loss": 4.3723,
+      "step": 1970
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019469809726859488,
+      "loss": 3.779,
+      "step": 1975
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019465787038899394,
+      "loss": 4.1662,
+      "step": 1980
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019461764350939297,
+      "loss": 3.2694,
+      "step": 1985
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019457741662979203,
+      "loss": 2.2375,
+      "step": 1990
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001945371897501911,
+      "loss": 2.9865,
+      "step": 1995
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019449696287059015,
+      "loss": 1.6985,
+      "step": 2000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019445673599098918,
+      "loss": 4.2811,
+      "step": 2005
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019441650911138824,
+      "loss": 4.726,
+      "step": 2010
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001943762822317873,
+      "loss": 4.2904,
+      "step": 2015
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019433605535218633,
+      "loss": 3.7736,
+      "step": 2020
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019429582847258538,
+      "loss": 3.3697,
+      "step": 2025
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019425560159298444,
+      "loss": 3.1179,
+      "step": 2030
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001942153747133835,
+      "loss": 2.9869,
+      "step": 2035
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019417514783378253,
+      "loss": 3.0633,
+      "step": 2040
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001941349209541816,
+      "loss": 3.2251,
+      "step": 2045
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019409469407458065,
+      "loss": 0.7779,
+      "step": 2050
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001940544671949797,
+      "loss": 4.3598,
+      "step": 2055
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019401424031537874,
+      "loss": 4.9221,
+      "step": 2060
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001939740134357778,
+      "loss": 4.2301,
+      "step": 2065
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019393378655617685,
+      "loss": 5.2469,
+      "step": 2070
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019389355967657589,
+      "loss": 3.3051,
+      "step": 2075
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019385333279697494,
+      "loss": 3.6274,
+      "step": 2080
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000193813105917374,
+      "loss": 4.6274,
+      "step": 2085
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019377287903777306,
+      "loss": 2.9797,
+      "step": 2090
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001937326521581721,
+      "loss": 2.6226,
+      "step": 2095
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019369242527857115,
+      "loss": 2.4438,
+      "step": 2100
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001936521983989702,
+      "loss": 4.2717,
+      "step": 2105
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019361197151936924,
+      "loss": 4.4154,
+      "step": 2110
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001935717446397683,
+      "loss": 5.1223,
+      "step": 2115
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019353151776016736,
+      "loss": 4.3166,
+      "step": 2120
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019349129088056641,
+      "loss": 3.5363,
+      "step": 2125
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019345106400096547,
+      "loss": 4.075,
+      "step": 2130
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001934108371213645,
+      "loss": 2.5993,
+      "step": 2135
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019337061024176356,
+      "loss": 2.8878,
+      "step": 2140
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001933303833621626,
+      "loss": 2.9542,
+      "step": 2145
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019329015648256165,
+      "loss": 3.0859,
+      "step": 2150
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001932499296029607,
+      "loss": 5.251,
+      "step": 2155
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019320970272335977,
+      "loss": 4.7705,
+      "step": 2160
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019316947584375883,
+      "loss": 4.5398,
+      "step": 2165
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019312924896415786,
+      "loss": 4.2553,
+      "step": 2170
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019308902208455691,
+      "loss": 5.5578,
+      "step": 2175
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019304879520495595,
+      "loss": 3.7627,
+      "step": 2180
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000193008568325355,
+      "loss": 3.8887,
+      "step": 2185
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019296834144575406,
+      "loss": 3.8691,
+      "step": 2190
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019292811456615312,
+      "loss": 2.8983,
+      "step": 2195
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019288788768655218,
+      "loss": 1.3964,
+      "step": 2200
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019284766080695124,
+      "loss": 5.0994,
+      "step": 2205
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019280743392735027,
+      "loss": 5.5555,
+      "step": 2210
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001927672070477493,
+      "loss": 4.6605,
+      "step": 2215
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019272698016814836,
+      "loss": 3.9756,
+      "step": 2220
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019268675328854742,
+      "loss": 4.1566,
+      "step": 2225
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019264652640894647,
+      "loss": 3.3558,
+      "step": 2230
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019260629952934553,
+      "loss": 4.5201,
+      "step": 2235
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001925660726497446,
+      "loss": 3.4633,
+      "step": 2240
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019252584577014362,
+      "loss": 3.0613,
+      "step": 2245
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019248561889054265,
+      "loss": 1.4768,
+      "step": 2250
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001924453920109417,
+      "loss": 5.6229,
+      "step": 2255
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019240516513134077,
+      "loss": 4.9016,
+      "step": 2260
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019236493825173983,
+      "loss": 4.9314,
+      "step": 2265
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019232471137213889,
+      "loss": 3.6584,
+      "step": 2270
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019228448449253794,
+      "loss": 2.9309,
+      "step": 2275
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019224425761293698,
+      "loss": 5.1328,
+      "step": 2280
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000192204030733336,
+      "loss": 1.9352,
+      "step": 2285
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019216380385373506,
+      "loss": 2.3981,
+      "step": 2290
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019212357697413412,
+      "loss": 4.707,
+      "step": 2295
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019208335009453318,
+      "loss": 0.5809,
+      "step": 2300
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019204312321493224,
+      "loss": 5.2449,
+      "step": 2305
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019200289633533127,
+      "loss": 4.7281,
+      "step": 2310
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019196266945573033,
+      "loss": 4.5684,
+      "step": 2315
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019192244257612936,
+      "loss": 4.249,
+      "step": 2320
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019188221569652842,
+      "loss": 3.5209,
+      "step": 2325
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019184198881692748,
+      "loss": 3.7265,
+      "step": 2330
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019180176193732653,
+      "loss": 3.7072,
+      "step": 2335
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001917615350577256,
+      "loss": 2.195,
+      "step": 2340
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019172130817812462,
+      "loss": 3.1261,
+      "step": 2345
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019168108129852368,
+      "loss": 1.413,
+      "step": 2350
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019164085441892274,
+      "loss": 4.6439,
+      "step": 2355
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019160062753932177,
+      "loss": 6.491,
+      "step": 2360
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019156040065972083,
+      "loss": 4.1846,
+      "step": 2365
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001915201737801199,
+      "loss": 4.8385,
+      "step": 2370
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019147994690051895,
+      "loss": 4.0875,
+      "step": 2375
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019143972002091798,
+      "loss": 3.4586,
+      "step": 2380
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019139949314131704,
+      "loss": 3.9514,
+      "step": 2385
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001913592662617161,
+      "loss": 1.9388,
+      "step": 2390
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019131903938211513,
+      "loss": 3.4899,
+      "step": 2395
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019127881250251418,
+      "loss": 0.8589,
+      "step": 2400
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019123858562291324,
+      "loss": 5.1305,
+      "step": 2405
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001911983587433123,
+      "loss": 5.6215,
+      "step": 2410
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019115813186371133,
+      "loss": 5.152,
+      "step": 2415
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001911179049841104,
+      "loss": 4.2031,
+      "step": 2420
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019107767810450945,
+      "loss": 3.8862,
+      "step": 2425
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001910374512249085,
+      "loss": 3.2398,
+      "step": 2430
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019099722434530754,
+      "loss": 3.3589,
+      "step": 2435
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001909569974657066,
+      "loss": 1.6699,
+      "step": 2440
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019091677058610565,
+      "loss": 2.6013,
+      "step": 2445
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019087654370650468,
+      "loss": 1.346,
+      "step": 2450
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019083631682690374,
+      "loss": 5.7719,
+      "step": 2455
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001907960899473028,
+      "loss": 4.8369,
+      "step": 2460
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019075586306770186,
+      "loss": 3.7029,
+      "step": 2465
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001907156361881009,
+      "loss": 2.6815,
+      "step": 2470
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019067540930849995,
+      "loss": 4.5318,
+      "step": 2475
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000190635182428899,
+      "loss": 4.4757,
+      "step": 2480
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019059495554929804,
+      "loss": 3.0317,
+      "step": 2485
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001905547286696971,
+      "loss": 3.3823,
+      "step": 2490
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019051450179009615,
+      "loss": 2.1039,
+      "step": 2495
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001904742749104952,
+      "loss": 1.3054,
+      "step": 2500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019043404803089427,
+      "loss": 4.5176,
+      "step": 2505
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001903938211512933,
+      "loss": 5.5738,
+      "step": 2510
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019035359427169236,
+      "loss": 4.4791,
+      "step": 2515
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001903133673920914,
+      "loss": 3.3446,
+      "step": 2520
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019027314051249045,
+      "loss": 3.6072,
+      "step": 2525
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001902329136328895,
+      "loss": 3.7518,
+      "step": 2530
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019019268675328857,
+      "loss": 3.9649,
+      "step": 2535
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019015245987368762,
+      "loss": 2.0974,
+      "step": 2540
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019011223299408666,
+      "loss": 1.7716,
+      "step": 2545
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019007200611448571,
+      "loss": 2.7839,
+      "step": 2550
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019003177923488474,
+      "loss": 5.41,
+      "step": 2555
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001899915523552838,
+      "loss": 4.9215,
+      "step": 2560
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018995132547568286,
+      "loss": 3.7002,
+      "step": 2565
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018991109859608192,
+      "loss": 4.8951,
+      "step": 2570
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018987087171648098,
+      "loss": 2.5297,
+      "step": 2575
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018983064483688004,
+      "loss": 2.9542,
+      "step": 2580
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018979041795727907,
+      "loss": 2.4865,
+      "step": 2585
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001897501910776781,
+      "loss": 2.3631,
+      "step": 2590
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018970996419807716,
+      "loss": 2.8949,
+      "step": 2595
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018966973731847621,
+      "loss": 0.6757,
+      "step": 2600
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018962951043887527,
+      "loss": 4.5512,
+      "step": 2605
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018958928355927433,
+      "loss": 5.1891,
+      "step": 2610
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018954905667967336,
+      "loss": 5.5706,
+      "step": 2615
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001895088298000724,
+      "loss": 6.3279,
+      "step": 2620
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018946860292047145,
+      "loss": 4.7213,
+      "step": 2625
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001894283760408705,
+      "loss": 3.6586,
+      "step": 2630
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018938814916126957,
+      "loss": 3.4731,
+      "step": 2635
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018934792228166863,
+      "loss": 2.9237,
+      "step": 2640
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018930769540206768,
+      "loss": 1.2635,
+      "step": 2645
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018926746852246672,
+      "loss": 2.0249,
+      "step": 2650
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018922724164286577,
+      "loss": 4.6764,
+      "step": 2655
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001891870147632648,
+      "loss": 5.0914,
+      "step": 2660
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018914678788366386,
+      "loss": 4.6342,
+      "step": 2665
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018910656100406292,
+      "loss": 4.3582,
+      "step": 2670
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018906633412446198,
+      "loss": 3.7621,
+      "step": 2675
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018902610724486104,
+      "loss": 3.5162,
+      "step": 2680
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018898588036526007,
+      "loss": 2.9408,
+      "step": 2685
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018894565348565913,
+      "loss": 2.3332,
+      "step": 2690
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018890542660605816,
+      "loss": 2.6293,
+      "step": 2695
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018886519972645722,
+      "loss": 0.7619,
+      "step": 2700
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018882497284685628,
+      "loss": 5.459,
+      "step": 2705
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018878474596725533,
+      "loss": 4.375,
+      "step": 2710
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001887445190876544,
+      "loss": 5.0764,
+      "step": 2715
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018870429220805342,
+      "loss": 3.6271,
+      "step": 2720
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018866406532845248,
+      "loss": 3.8543,
+      "step": 2725
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018862383844885154,
+      "loss": 2.9905,
+      "step": 2730
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018858361156925057,
+      "loss": 3.3131,
+      "step": 2735
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018854338468964963,
+      "loss": 2.6086,
+      "step": 2740
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001885031578100487,
+      "loss": 1.4445,
+      "step": 2745
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018846293093044775,
+      "loss": 1.3326,
+      "step": 2750
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018842270405084678,
+      "loss": 3.508,
+      "step": 2755
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018838247717124583,
+      "loss": 4.6805,
+      "step": 2760
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001883422502916449,
+      "loss": 5.0881,
+      "step": 2765
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018830202341204392,
+      "loss": 4.0855,
+      "step": 2770
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018826179653244298,
+      "loss": 2.9991,
+      "step": 2775
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018822156965284204,
+      "loss": 2.9354,
+      "step": 2780
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001881813427732411,
+      "loss": 1.5435,
+      "step": 2785
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018814111589364013,
+      "loss": 2.5249,
+      "step": 2790
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001881008890140392,
+      "loss": 1.2265,
+      "step": 2795
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018806066213443825,
+      "loss": 0.833,
+      "step": 2800
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001880204352548373,
+      "loss": 5.0906,
+      "step": 2805
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018798020837523634,
+      "loss": 5.0311,
+      "step": 2810
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001879399814956354,
+      "loss": 5.8641,
+      "step": 2815
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018789975461603445,
+      "loss": 4.1654,
+      "step": 2820
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018785952773643348,
+      "loss": 3.4861,
+      "step": 2825
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018781930085683254,
+      "loss": 3.3804,
+      "step": 2830
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001877790739772316,
+      "loss": 4.5379,
+      "step": 2835
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018773884709763066,
+      "loss": 1.8242,
+      "step": 2840
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001876986202180297,
+      "loss": 3.9177,
+      "step": 2845
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018765839333842875,
+      "loss": 0.7359,
+      "step": 2850
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001876181664588278,
+      "loss": 6.2848,
+      "step": 2855
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018757793957922684,
+      "loss": 4.1291,
+      "step": 2860
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001875377126996259,
+      "loss": 4.0441,
+      "step": 2865
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018749748582002495,
+      "loss": 3.3439,
+      "step": 2870
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000187457258940424,
+      "loss": 3.2429,
+      "step": 2875
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018741703206082307,
+      "loss": 3.4903,
+      "step": 2880
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001873768051812221,
+      "loss": 3.8007,
+      "step": 2885
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018733657830162116,
+      "loss": 2.9526,
+      "step": 2890
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001872963514220202,
+      "loss": 3.0772,
+      "step": 2895
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018725612454241925,
+      "loss": 0.8114,
+      "step": 2900
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001872158976628183,
+      "loss": 5.1357,
+      "step": 2905
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018717567078321737,
+      "loss": 5.2641,
+      "step": 2910
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018713544390361642,
+      "loss": 4.9559,
+      "step": 2915
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018709521702401545,
+      "loss": 4.3744,
+      "step": 2920
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018705499014441449,
+      "loss": 4.2383,
+      "step": 2925
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018701476326481354,
+      "loss": 3.8998,
+      "step": 2930
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001869745363852126,
+      "loss": 2.7238,
+      "step": 2935
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018693430950561166,
+      "loss": 2.415,
+      "step": 2940
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018689408262601072,
+      "loss": 2.7549,
+      "step": 2945
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018685385574640978,
+      "loss": 2.9196,
+      "step": 2950
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001868136288668088,
+      "loss": 4.4637,
+      "step": 2955
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018677340198720784,
+      "loss": 4.441,
+      "step": 2960
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001867331751076069,
+      "loss": 4.4455,
+      "step": 2965
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018669294822800596,
+      "loss": 3.5803,
+      "step": 2970
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018665272134840501,
+      "loss": 3.7873,
+      "step": 2975
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018661249446880407,
+      "loss": 4.2402,
+      "step": 2980
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018657226758920313,
+      "loss": 4.591,
+      "step": 2985
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018653204070960216,
+      "loss": 2.1384,
+      "step": 2990
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001864918138300012,
+      "loss": 1.5015,
+      "step": 2995
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018645158695040025,
+      "loss": 1.4467,
+      "step": 3000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001864113600707993,
+      "loss": 5.4559,
+      "step": 3005
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018637113319119837,
+      "loss": 4.7643,
+      "step": 3010
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018633090631159743,
+      "loss": 5.4004,
+      "step": 3015
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018629067943199648,
+      "loss": 3.9115,
+      "step": 3020
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018625045255239552,
+      "loss": 3.2608,
+      "step": 3025
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018621022567279457,
+      "loss": 3.4641,
+      "step": 3030
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001861699987931936,
+      "loss": 3.2434,
+      "step": 3035
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018612977191359266,
+      "loss": 2.1317,
+      "step": 3040
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018608954503399172,
+      "loss": 2.0085,
+      "step": 3045
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018604931815439078,
+      "loss": 0.4546,
+      "step": 3050
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018600909127478984,
+      "loss": 4.4299,
+      "step": 3055
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018596886439518887,
+      "loss": 5.439,
+      "step": 3060
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018592863751558793,
+      "loss": 5.1666,
+      "step": 3065
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018588841063598696,
+      "loss": 4.3708,
+      "step": 3070
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018584818375638602,
+      "loss": 4.1246,
+      "step": 3075
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018580795687678507,
+      "loss": 4.082,
+      "step": 3080
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018576772999718413,
+      "loss": 2.4525,
+      "step": 3085
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001857275031175832,
+      "loss": 3.7855,
+      "step": 3090
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018568727623798222,
+      "loss": 1.7568,
+      "step": 3095
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018564704935838128,
+      "loss": 1.5382,
+      "step": 3100
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018560682247878034,
+      "loss": 4.3197,
+      "step": 3105
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018556659559917937,
+      "loss": 5.0633,
+      "step": 3110
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018552636871957843,
+      "loss": 4.0307,
+      "step": 3115
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018548614183997749,
+      "loss": 3.3434,
+      "step": 3120
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018544591496037654,
+      "loss": 4.3059,
+      "step": 3125
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018540568808077558,
+      "loss": 4.701,
+      "step": 3130
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018536546120117463,
+      "loss": 2.9802,
+      "step": 3135
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001853252343215737,
+      "loss": 1.449,
+      "step": 3140
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018528500744197272,
+      "loss": 3.2532,
+      "step": 3145
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018524478056237178,
+      "loss": 0.6617,
+      "step": 3150
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018520455368277084,
+      "loss": 4.2574,
+      "step": 3155
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001851643268031699,
+      "loss": 4.36,
+      "step": 3160
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018512409992356893,
+      "loss": 4.2436,
+      "step": 3165
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000185083873043968,
+      "loss": 4.8123,
+      "step": 3170
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018504364616436705,
+      "loss": 5.1578,
+      "step": 3175
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001850034192847661,
+      "loss": 4.4125,
+      "step": 3180
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018496319240516513,
+      "loss": 2.5215,
+      "step": 3185
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001849229655255642,
+      "loss": 3.3153,
+      "step": 3190
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018488273864596325,
+      "loss": 3.2267,
+      "step": 3195
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018484251176636228,
+      "loss": 2.41,
+      "step": 3200
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018480228488676134,
+      "loss": 5.0113,
+      "step": 3205
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001847620580071604,
+      "loss": 4.7652,
+      "step": 3210
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018472183112755946,
+      "loss": 4.9766,
+      "step": 3215
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001846816042479585,
+      "loss": 4.8104,
+      "step": 3220
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018464137736835755,
+      "loss": 4.0664,
+      "step": 3225
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018460115048875658,
+      "loss": 3.1479,
+      "step": 3230
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018456092360915564,
+      "loss": 3.6193,
+      "step": 3235
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001845206967295547,
+      "loss": 2.3463,
+      "step": 3240
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018448046984995375,
+      "loss": 2.5716,
+      "step": 3245
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001844402429703528,
+      "loss": 1.1213,
+      "step": 3250
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018440001609075187,
+      "loss": 5.2396,
+      "step": 3255
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001843597892111509,
+      "loss": 4.9785,
+      "step": 3260
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018431956233154993,
+      "loss": 4.757,
+      "step": 3265
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000184279335451949,
+      "loss": 4.6502,
+      "step": 3270
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018423910857234805,
+      "loss": 4.1502,
+      "step": 3275
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001841988816927471,
+      "loss": 3.6057,
+      "step": 3280
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018415865481314616,
+      "loss": 4.0316,
+      "step": 3285
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018411842793354522,
+      "loss": 3.9748,
+      "step": 3290
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018407820105394425,
+      "loss": 2.0564,
+      "step": 3295
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018403797417434328,
+      "loss": 1.9967,
+      "step": 3300
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018399774729474234,
+      "loss": 5.2789,
+      "step": 3305
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001839575204151414,
+      "loss": 4.1012,
+      "step": 3310
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018391729353554046,
+      "loss": 4.8074,
+      "step": 3315
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018387706665593952,
+      "loss": 2.8919,
+      "step": 3320
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018383683977633858,
+      "loss": 3.7766,
+      "step": 3325
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001837966128967376,
+      "loss": 5.0438,
+      "step": 3330
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018375638601713664,
+      "loss": 2.7931,
+      "step": 3335
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001837161591375357,
+      "loss": 2.0797,
+      "step": 3340
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018367593225793475,
+      "loss": 2.7992,
+      "step": 3345
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001836357053783338,
+      "loss": 0.7685,
+      "step": 3350
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018359547849873287,
+      "loss": 4.7795,
+      "step": 3355
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018355525161913193,
+      "loss": 4.2709,
+      "step": 3360
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018351502473953096,
+      "loss": 5.034,
+      "step": 3365
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018347479785993,
+      "loss": 4.5508,
+      "step": 3370
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018343457098032905,
+      "loss": 3.427,
+      "step": 3375
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001833943441007281,
+      "loss": 3.2379,
+      "step": 3380
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018335411722112717,
+      "loss": 3.8982,
+      "step": 3385
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018331389034152622,
+      "loss": 2.4782,
+      "step": 3390
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018327366346192528,
+      "loss": 3.1309,
+      "step": 3395
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018323343658232431,
+      "loss": 1.8897,
+      "step": 3400
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018319320970272337,
+      "loss": 5.2643,
+      "step": 3405
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001831529828231224,
+      "loss": 4.215,
+      "step": 3410
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018311275594352146,
+      "loss": 4.032,
+      "step": 3415
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018307252906392052,
+      "loss": 3.9859,
+      "step": 3420
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018303230218431958,
+      "loss": 3.4547,
+      "step": 3425
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018299207530471864,
+      "loss": 3.0241,
+      "step": 3430
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018295184842511767,
+      "loss": 2.1052,
+      "step": 3435
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018291162154551673,
+      "loss": 1.5382,
+      "step": 3440
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018287139466591576,
+      "loss": 4.3531,
+      "step": 3445
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018283116778631482,
+      "loss": 1.5439,
+      "step": 3450
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018279094090671387,
+      "loss": 4.5682,
+      "step": 3455
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018275071402711293,
+      "loss": 4.4668,
+      "step": 3460
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000182710487147512,
+      "loss": 5.041,
+      "step": 3465
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018267026026791102,
+      "loss": 4.1639,
+      "step": 3470
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018263003338831008,
+      "loss": 4.3855,
+      "step": 3475
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018258980650870914,
+      "loss": 3.7016,
+      "step": 3480
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018254957962910817,
+      "loss": 3.9585,
+      "step": 3485
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018250935274950723,
+      "loss": 2.0294,
+      "step": 3490
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018246912586990629,
+      "loss": 2.6052,
+      "step": 3495
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018242889899030534,
+      "loss": 1.1796,
+      "step": 3500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018238867211070437,
+      "loss": 3.7271,
+      "step": 3505
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018234844523110343,
+      "loss": 4.1547,
+      "step": 3510
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001823082183515025,
+      "loss": 4.6098,
+      "step": 3515
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018226799147190152,
+      "loss": 5.2176,
+      "step": 3520
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018222776459230058,
+      "loss": 4.0721,
+      "step": 3525
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018218753771269964,
+      "loss": 4.3234,
+      "step": 3530
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018214731083309867,
+      "loss": 3.1592,
+      "step": 3535
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018210708395349773,
+      "loss": 3.2439,
+      "step": 3540
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018206685707389679,
+      "loss": 2.8346,
+      "step": 3545
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018202663019429584,
+      "loss": 1.0432,
+      "step": 3550
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001819864033146949,
+      "loss": 4.7693,
+      "step": 3555
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018194617643509393,
+      "loss": 3.7051,
+      "step": 3560
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000181905949555493,
+      "loss": 4.0299,
+      "step": 3565
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018186572267589202,
+      "loss": 4.5971,
+      "step": 3570
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018182549579629108,
+      "loss": 3.5709,
+      "step": 3575
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018178526891669014,
+      "loss": 3.5243,
+      "step": 3580
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001817450420370892,
+      "loss": 2.8105,
+      "step": 3585
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018170481515748826,
+      "loss": 1.6252,
+      "step": 3590
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001816645882778873,
+      "loss": 1.5084,
+      "step": 3595
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018162436139828635,
+      "loss": 1.4158,
+      "step": 3600
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018158413451868538,
+      "loss": 4.3527,
+      "step": 3605
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018154390763908444,
+      "loss": 6.2471,
+      "step": 3610
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001815036807594835,
+      "loss": 3.4996,
+      "step": 3615
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018146345387988255,
+      "loss": 3.4791,
+      "step": 3620
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001814232270002816,
+      "loss": 3.9434,
+      "step": 3625
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018138300012068067,
+      "loss": 3.8335,
+      "step": 3630
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001813427732410797,
+      "loss": 2.7479,
+      "step": 3635
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018130254636147873,
+      "loss": 3.481,
+      "step": 3640
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001812623194818778,
+      "loss": 2.2178,
+      "step": 3645
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018122209260227685,
+      "loss": 0.5455,
+      "step": 3650
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001811818657226759,
+      "loss": 4.8406,
+      "step": 3655
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018114163884307496,
+      "loss": 4.449,
+      "step": 3660
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018110141196347402,
+      "loss": 5.3301,
+      "step": 3665
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018106118508387305,
+      "loss": 5.1273,
+      "step": 3670
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018102095820427208,
+      "loss": 3.4895,
+      "step": 3675
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018098073132467114,
+      "loss": 2.6304,
+      "step": 3680
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001809405044450702,
+      "loss": 2.2214,
+      "step": 3685
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018090027756546926,
+      "loss": 2.3575,
+      "step": 3690
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018086005068586832,
+      "loss": 1.2417,
+      "step": 3695
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018081982380626738,
+      "loss": 2.804,
+      "step": 3700
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001807795969266664,
+      "loss": 4.5449,
+      "step": 3705
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018073937004706544,
+      "loss": 4.7197,
+      "step": 3710
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001806991431674645,
+      "loss": 5.1338,
+      "step": 3715
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018065891628786355,
+      "loss": 4.2385,
+      "step": 3720
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001806186894082626,
+      "loss": 3.7326,
+      "step": 3725
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00018057846252866167,
+      "loss": 3.5176,
+      "step": 3730
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00018053823564906073,
+      "loss": 2.6052,
+      "step": 3735
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00018049800876945976,
+      "loss": 2.286,
+      "step": 3740
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001804577818898588,
+      "loss": 3.0982,
+      "step": 3745
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00018041755501025785,
+      "loss": 2.8813,
+      "step": 3750
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001803773281306569,
+      "loss": 4.1352,
+      "step": 3755
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00018033710125105597,
+      "loss": 4.5643,
+      "step": 3760
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00018029687437145502,
+      "loss": 4.6842,
+      "step": 3765
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00018025664749185408,
+      "loss": 3.4552,
+      "step": 3770
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001802164206122531,
+      "loss": 3.9146,
+      "step": 3775
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00018017619373265217,
+      "loss": 3.2099,
+      "step": 3780
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001801359668530512,
+      "loss": 4.366,
+      "step": 3785
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00018009573997345026,
+      "loss": 2.5956,
+      "step": 3790
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00018005551309384932,
+      "loss": 1.6513,
+      "step": 3795
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00018001528621424838,
+      "loss": 2.3943,
+      "step": 3800
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017997505933464744,
+      "loss": 4.4051,
+      "step": 3805
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017993483245504647,
+      "loss": 4.1432,
+      "step": 3810
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017989460557544552,
+      "loss": 4.6814,
+      "step": 3815
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017985437869584456,
+      "loss": 3.7793,
+      "step": 3820
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017981415181624361,
+      "loss": 3.6339,
+      "step": 3825
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017977392493664267,
+      "loss": 3.716,
+      "step": 3830
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017973369805704173,
+      "loss": 3.7518,
+      "step": 3835
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017969347117744076,
+      "loss": 3.3349,
+      "step": 3840
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017965324429783982,
+      "loss": 1.3231,
+      "step": 3845
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017961301741823888,
+      "loss": 2.5204,
+      "step": 3850
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017957279053863794,
+      "loss": 5.1436,
+      "step": 3855
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017953256365903697,
+      "loss": 5.2109,
+      "step": 3860
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017949233677943603,
+      "loss": 3.9754,
+      "step": 3865
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017945210989983508,
+      "loss": 3.8529,
+      "step": 3870
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017941188302023412,
+      "loss": 4.5021,
+      "step": 3875
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017937165614063317,
+      "loss": 3.8877,
+      "step": 3880
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017933142926103223,
+      "loss": 1.8866,
+      "step": 3885
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001792912023814313,
+      "loss": 4.2512,
+      "step": 3890
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017925097550183032,
+      "loss": 2.6688,
+      "step": 3895
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017921074862222938,
+      "loss": 3.2995,
+      "step": 3900
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017917052174262844,
+      "loss": 4.2304,
+      "step": 3905
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017913029486302747,
+      "loss": 4.9676,
+      "step": 3910
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017909006798342653,
+      "loss": 4.2441,
+      "step": 3915
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017904984110382559,
+      "loss": 3.8947,
+      "step": 3920
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017900961422422464,
+      "loss": 4.1439,
+      "step": 3925
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001789693873446237,
+      "loss": 3.5044,
+      "step": 3930
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017892916046502273,
+      "loss": 2.705,
+      "step": 3935
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001788889335854218,
+      "loss": 0.9807,
+      "step": 3940
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017884870670582082,
+      "loss": 2.8294,
+      "step": 3945
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017880847982621988,
+      "loss": 0.8584,
+      "step": 3950
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017876825294661894,
+      "loss": 4.865,
+      "step": 3955
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000178728026067018,
+      "loss": 4.4303,
+      "step": 3960
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017868779918741706,
+      "loss": 4.9551,
+      "step": 3965
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017864757230781611,
+      "loss": 4.6406,
+      "step": 3970
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017860734542821514,
+      "loss": 3.2033,
+      "step": 3975
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017856711854861418,
+      "loss": 3.6012,
+      "step": 3980
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017852689166901323,
+      "loss": 1.4554,
+      "step": 3985
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001784866647894123,
+      "loss": 2.4178,
+      "step": 3990
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017844643790981135,
+      "loss": 1.7368,
+      "step": 3995
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001784062110302104,
+      "loss": 1.9992,
+      "step": 4000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017836598415060947,
+      "loss": 3.7744,
+      "step": 4005
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001783257572710085,
+      "loss": 4.1831,
+      "step": 4010
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017828553039140753,
+      "loss": 3.9318,
+      "step": 4015
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001782453035118066,
+      "loss": 3.6878,
+      "step": 4020
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017820507663220565,
+      "loss": 4.0703,
+      "step": 4025
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001781648497526047,
+      "loss": 3.9203,
+      "step": 4030
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017812462287300376,
+      "loss": 2.88,
+      "step": 4035
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017808439599340282,
+      "loss": 2.49,
+      "step": 4040
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017804416911380185,
+      "loss": 1.8029,
+      "step": 4045
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017800394223420088,
+      "loss": 1.4607,
+      "step": 4050
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017796371535459994,
+      "loss": 4.866,
+      "step": 4055
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000177923488474999,
+      "loss": 4.9695,
+      "step": 4060
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017788326159539806,
+      "loss": 4.073,
+      "step": 4065
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017784303471579712,
+      "loss": 4.3596,
+      "step": 4070
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017780280783619617,
+      "loss": 3.5149,
+      "step": 4075
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001777625809565952,
+      "loss": 3.9059,
+      "step": 4080
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017772235407699424,
+      "loss": 2.6699,
+      "step": 4085
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001776821271973933,
+      "loss": 3.2956,
+      "step": 4090
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017764190031779235,
+      "loss": 2.8765,
+      "step": 4095
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001776016734381914,
+      "loss": 2.6877,
+      "step": 4100
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017756144655859047,
+      "loss": 4.8078,
+      "step": 4105
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017752121967898953,
+      "loss": 4.3077,
+      "step": 4110
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017748099279938856,
+      "loss": 3.7305,
+      "step": 4115
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017744076591978762,
+      "loss": 3.5543,
+      "step": 4120
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017740053904018665,
+      "loss": 4.3678,
+      "step": 4125
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001773603121605857,
+      "loss": 3.1074,
+      "step": 4130
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017732008528098476,
+      "loss": 3.485,
+      "step": 4135
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017727985840138382,
+      "loss": 2.4198,
+      "step": 4140
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017723963152178285,
+      "loss": 2.7068,
+      "step": 4145
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001771994046421819,
+      "loss": 2.6594,
+      "step": 4150
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017715917776258097,
+      "loss": 3.9896,
+      "step": 4155
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017711895088298,
+      "loss": 4.7062,
+      "step": 4160
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017707872400337906,
+      "loss": 5.0463,
+      "step": 4165
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017703849712377812,
+      "loss": 3.9547,
+      "step": 4170
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017699827024417718,
+      "loss": 4.6994,
+      "step": 4175
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001769580433645762,
+      "loss": 4.1545,
+      "step": 4180
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017691781648497527,
+      "loss": 3.0788,
+      "step": 4185
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017687758960537432,
+      "loss": 2.744,
+      "step": 4190
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017683736272577338,
+      "loss": 2.6146,
+      "step": 4195
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017680518122209263,
+      "loss": 3.9939,
+      "step": 4200
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017676495434249166,
+      "loss": 4.5344,
+      "step": 4205
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001767247274628907,
+      "loss": 4.7307,
+      "step": 4210
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017668450058328975,
+      "loss": 4.5513,
+      "step": 4215
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001766442737036888,
+      "loss": 3.2362,
+      "step": 4220
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017660404682408787,
+      "loss": 3.5146,
+      "step": 4225
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017656381994448692,
+      "loss": 2.4402,
+      "step": 4230
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017652359306488598,
+      "loss": 2.7357,
+      "step": 4235
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000176483366185285,
+      "loss": 2.2243,
+      "step": 4240
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017644313930568404,
+      "loss": 2.6992,
+      "step": 4245
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001764029124260831,
+      "loss": 2.3734,
+      "step": 4250
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017636268554648216,
+      "loss": 4.1707,
+      "step": 4255
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017632245866688122,
+      "loss": 5.4875,
+      "step": 4260
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017628223178728028,
+      "loss": 4.6299,
+      "step": 4265
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017624200490767934,
+      "loss": 3.0701,
+      "step": 4270
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017620177802807837,
+      "loss": 3.775,
+      "step": 4275
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001761615511484774,
+      "loss": 2.9104,
+      "step": 4280
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017612132426887646,
+      "loss": 2.7511,
+      "step": 4285
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017608109738927551,
+      "loss": 2.603,
+      "step": 4290
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017604087050967457,
+      "loss": 2.2994,
+      "step": 4295
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017600064363007363,
+      "loss": 2.7078,
+      "step": 4300
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001759604167504727,
+      "loss": 4.4588,
+      "step": 4305
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017592018987087172,
+      "loss": 4.8812,
+      "step": 4310
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017587996299127078,
+      "loss": 4.8076,
+      "step": 4315
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001758397361116698,
+      "loss": 3.1662,
+      "step": 4320
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017579950923206887,
+      "loss": 4.4642,
+      "step": 4325
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017575928235246793,
+      "loss": 2.0759,
+      "step": 4330
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017571905547286698,
+      "loss": 2.2219,
+      "step": 4335
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017567882859326604,
+      "loss": 1.4699,
+      "step": 4340
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017563860171366507,
+      "loss": 1.0181,
+      "step": 4345
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017559837483406413,
+      "loss": 1.4557,
+      "step": 4350
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017555814795446316,
+      "loss": 5.0143,
+      "step": 4355
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017551792107486222,
+      "loss": 4.6441,
+      "step": 4360
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017547769419526128,
+      "loss": 4.8084,
+      "step": 4365
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017543746731566034,
+      "loss": 5.1527,
+      "step": 4370
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001753972404360594,
+      "loss": 4.5236,
+      "step": 4375
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017535701355645843,
+      "loss": 4.1115,
+      "step": 4380
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017531678667685749,
+      "loss": 3.2589,
+      "step": 4385
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017527655979725654,
+      "loss": 3.387,
+      "step": 4390
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017523633291765558,
+      "loss": 1.2526,
+      "step": 4395
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017519610603805463,
+      "loss": 0.799,
+      "step": 4400
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001751558791584537,
+      "loss": 5.3492,
+      "step": 4405
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017511565227885275,
+      "loss": 4.1848,
+      "step": 4410
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017507542539925178,
+      "loss": 5.2502,
+      "step": 4415
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017503519851965084,
+      "loss": 4.8549,
+      "step": 4420
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001749949716400499,
+      "loss": 4.3082,
+      "step": 4425
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017495474476044893,
+      "loss": 3.2034,
+      "step": 4430
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000174914517880848,
+      "loss": 3.5008,
+      "step": 4435
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017487429100124705,
+      "loss": 2.2585,
+      "step": 4440
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001748340641216461,
+      "loss": 1.9539,
+      "step": 4445
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017479383724204513,
+      "loss": 1.5011,
+      "step": 4450
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001747536103624442,
+      "loss": 4.0939,
+      "step": 4455
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017471338348284325,
+      "loss": 4.458,
+      "step": 4460
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001746731566032423,
+      "loss": 6.1611,
+      "step": 4465
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017463292972364134,
+      "loss": 4.3582,
+      "step": 4470
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001745927028440404,
+      "loss": 3.2851,
+      "step": 4475
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017455247596443946,
+      "loss": 2.6692,
+      "step": 4480
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001745122490848385,
+      "loss": 3.419,
+      "step": 4485
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017447202220523755,
+      "loss": 3.178,
+      "step": 4490
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001744317953256366,
+      "loss": 2.9248,
+      "step": 4495
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017439156844603566,
+      "loss": 2.8947,
+      "step": 4500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001743513415664347,
+      "loss": 5.7676,
+      "step": 4505
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017431111468683375,
+      "loss": 4.4447,
+      "step": 4510
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00017427088780723278,
+      "loss": 4.7262,
+      "step": 4515
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00017423066092763184,
+      "loss": 3.8962,
+      "step": 4520
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001741904340480309,
+      "loss": 3.9734,
+      "step": 4525
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00017415020716842996,
+      "loss": 3.832,
+      "step": 4530
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00017410998028882902,
+      "loss": 2.866,
+      "step": 4535
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00017406975340922807,
+      "loss": 2.5358,
+      "step": 4540
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001740295265296271,
+      "loss": 1.7431,
+      "step": 4545
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00017398929965002614,
+      "loss": 2.6781,
+      "step": 4550
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001739490727704252,
+      "loss": 4.4768,
+      "step": 4555
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00017390884589082425,
+      "loss": 3.8754,
+      "step": 4560
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001738686190112233,
+      "loss": 3.7852,
+      "step": 4565
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00017382839213162237,
+      "loss": 4.2292,
+      "step": 4570
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00017378816525202143,
+      "loss": 3.6535,
+      "step": 4575
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00017374793837242046,
+      "loss": 4.4597,
+      "step": 4580
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001737077114928195,
+      "loss": 2.1313,
+      "step": 4585
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00017366748461321855,
+      "loss": 2.3775,
+      "step": 4590
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001736272577336176,
+      "loss": 1.5729,
+      "step": 4595
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00017358703085401666,
+      "loss": 1.3116,
+      "step": 4600
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00017354680397441572,
+      "loss": 5.5291,
+      "step": 4605
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00017350657709481478,
+      "loss": 4.3365,
+      "step": 4610
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001734663502152138,
+      "loss": 3.6343,
+      "step": 4615
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00017342612333561284,
+      "loss": 4.5385,
+      "step": 4620
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001733858964560119,
+      "loss": 3.4747,
+      "step": 4625
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00017334566957641096,
+      "loss": 2.505,
+      "step": 4630
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00017330544269681002,
+      "loss": 2.2642,
+      "step": 4635
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00017326521581720908,
+      "loss": 1.7137,
+      "step": 4640
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00017322498893760813,
+      "loss": 2.2715,
+      "step": 4645
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00017318476205800717,
+      "loss": 2.6712,
+      "step": 4650
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001731445351784062,
+      "loss": 4.7334,
+      "step": 4655
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00017310430829880526,
+      "loss": 3.4988,
+      "step": 4660
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00017306408141920431,
+      "loss": 3.4859,
+      "step": 4665
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00017302385453960337,
+      "loss": 2.9528,
+      "step": 4670
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00017298362766000243,
+      "loss": 4.2112,
+      "step": 4675
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001729434007804015,
+      "loss": 4.345,
+      "step": 4680
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00017290317390080052,
+      "loss": 2.7534,
+      "step": 4685
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00017286294702119958,
+      "loss": 1.9547,
+      "step": 4690
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001728227201415986,
+      "loss": 2.9516,
+      "step": 4695
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00017278249326199767,
+      "loss": 0.7848,
+      "step": 4700
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00017274226638239673,
+      "loss": 5.2662,
+      "step": 4705
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00017270203950279578,
+      "loss": 5.1359,
+      "step": 4710
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00017266181262319484,
+      "loss": 4.5707,
+      "step": 4715
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00017262158574359387,
+      "loss": 3.8627,
+      "step": 4720
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00017258135886399293,
+      "loss": 5.6158,
+      "step": 4725
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00017254113198439196,
+      "loss": 4.3295,
+      "step": 4730
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00017250090510479102,
+      "loss": 2.5796,
+      "step": 4735
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00017246067822519008,
+      "loss": 2.2212,
+      "step": 4740
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00017242045134558914,
+      "loss": 1.7833,
+      "step": 4745
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001723802244659882,
+      "loss": 1.0688,
+      "step": 4750
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00017233999758638723,
+      "loss": 5.7943,
+      "step": 4755
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00017229977070678628,
+      "loss": 4.3023,
+      "step": 4760
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00017225954382718534,
+      "loss": 4.1447,
+      "step": 4765
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00017221931694758437,
+      "loss": 4.5957,
+      "step": 4770
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00017217909006798343,
+      "loss": 4.5724,
+      "step": 4775
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001721388631883825,
+      "loss": 3.3936,
+      "step": 4780
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00017209863630878155,
+      "loss": 1.8421,
+      "step": 4785
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00017205840942918058,
+      "loss": 2.0833,
+      "step": 4790
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00017201818254957964,
+      "loss": 2.1733,
+      "step": 4795
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001719779556699787,
+      "loss": 1.4903,
+      "step": 4800
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00017193772879037773,
+      "loss": 5.8072,
+      "step": 4805
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00017189750191077679,
+      "loss": 4.2612,
+      "step": 4810
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00017185727503117584,
+      "loss": 3.9973,
+      "step": 4815
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00017181704815157488,
+      "loss": 4.2791,
+      "step": 4820
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00017177682127197393,
+      "loss": 4.2564,
+      "step": 4825
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000171736594392373,
+      "loss": 4.493,
+      "step": 4830
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00017169636751277205,
+      "loss": 3.7284,
+      "step": 4835
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001716561406331711,
+      "loss": 2.7577,
+      "step": 4840
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00017161591375357014,
+      "loss": 1.5587,
+      "step": 4845
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001715756868739692,
+      "loss": 1.974,
+      "step": 4850
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00017153545999436823,
+      "loss": 4.9592,
+      "step": 4855
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001714952331147673,
+      "loss": 3.9775,
+      "step": 4860
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00017145500623516635,
+      "loss": 4.7291,
+      "step": 4865
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001714147793555654,
+      "loss": 4.1361,
+      "step": 4870
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00017137455247596446,
+      "loss": 4.0131,
+      "step": 4875
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001713343255963635,
+      "loss": 4.918,
+      "step": 4880
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00017129409871676255,
+      "loss": 3.039,
+      "step": 4885
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00017125387183716158,
+      "loss": 2.5513,
+      "step": 4890
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00017121364495756064,
+      "loss": 1.4615,
+      "step": 4895
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001711734180779597,
+      "loss": 1.9534,
+      "step": 4900
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00017113319119835876,
+      "loss": 4.6439,
+      "step": 4905
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00017109296431875782,
+      "loss": 4.8441,
+      "step": 4910
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00017105273743915687,
+      "loss": 3.5526,
+      "step": 4915
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001710125105595559,
+      "loss": 3.2357,
+      "step": 4920
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00017097228367995494,
+      "loss": 3.4221,
+      "step": 4925
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000170932056800354,
+      "loss": 3.1509,
+      "step": 4930
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00017089182992075305,
+      "loss": 1.6275,
+      "step": 4935
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001708516030411521,
+      "loss": 2.1762,
+      "step": 4940
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00017081137616155117,
+      "loss": 1.9439,
+      "step": 4945
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00017077114928195023,
+      "loss": 0.8666,
+      "step": 4950
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00017073092240234926,
+      "loss": 5.1297,
+      "step": 4955
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001706906955227483,
+      "loss": 4.4209,
+      "step": 4960
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00017065046864314735,
+      "loss": 4.9455,
+      "step": 4965
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001706102417635464,
+      "loss": 4.8725,
+      "step": 4970
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00017057001488394546,
+      "loss": 3.9418,
+      "step": 4975
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00017052978800434452,
+      "loss": 4.1444,
+      "step": 4980
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00017048956112474358,
+      "loss": 3.5028,
+      "step": 4985
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001704493342451426,
+      "loss": 2.5432,
+      "step": 4990
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00017040910736554164,
+      "loss": 1.4747,
+      "step": 4995
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001703688804859407,
+      "loss": 0.9428,
+      "step": 5000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00017032865360633976,
+      "loss": 5.8418,
+      "step": 5005
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00017028842672673882,
+      "loss": 3.3178,
+      "step": 5010
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00017024819984713788,
+      "loss": 4.2133,
+      "step": 5015
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00017020797296753693,
+      "loss": 3.9502,
+      "step": 5020
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00017016774608793597,
+      "loss": 3.731,
+      "step": 5025
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000170127519208335,
+      "loss": 4.6752,
+      "step": 5030
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00017008729232873405,
+      "loss": 2.7182,
+      "step": 5035
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001700470654491331,
+      "loss": 2.9108,
+      "step": 5040
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00017000683856953217,
+      "loss": 3.8828,
+      "step": 5045
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00016996661168993123,
+      "loss": 1.53,
+      "step": 5050
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001699263848103303,
+      "loss": 5.5035,
+      "step": 5055
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00016988615793072932,
+      "loss": 3.9553,
+      "step": 5060
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00016984593105112838,
+      "loss": 4.5656,
+      "step": 5065
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001698057041715274,
+      "loss": 4.4553,
+      "step": 5070
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00016976547729192647,
+      "loss": 4.3305,
+      "step": 5075
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00016972525041232552,
+      "loss": 3.2616,
+      "step": 5080
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00016968502353272458,
+      "loss": 4.177,
+      "step": 5085
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00016964479665312364,
+      "loss": 3.3635,
+      "step": 5090
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00016960456977352267,
+      "loss": 0.7162,
+      "step": 5095
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00016956434289392173,
+      "loss": 3.3646,
+      "step": 5100
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00016952411601432076,
+      "loss": 3.9,
+      "step": 5105
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00016948388913471982,
+      "loss": 5.1976,
+      "step": 5110
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00016944366225511888,
+      "loss": 5.4234,
+      "step": 5115
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00016940343537551794,
+      "loss": 3.2449,
+      "step": 5120
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00016936320849591697,
+      "loss": 3.8414,
+      "step": 5125
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00016932298161631603,
+      "loss": 3.5964,
+      "step": 5130
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00016928275473671508,
+      "loss": 3.0642,
+      "step": 5135
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00016924252785711414,
+      "loss": 2.8784,
+      "step": 5140
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00016920230097751317,
+      "loss": 2.0424,
+      "step": 5145
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00016916207409791223,
+      "loss": 0.0858,
+      "step": 5150
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001691218472183113,
+      "loss": 4.6318,
+      "step": 5155
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00016908162033871032,
+      "loss": 5.0166,
+      "step": 5160
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00016904139345910938,
+      "loss": 3.9994,
+      "step": 5165
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00016900116657950844,
+      "loss": 4.776,
+      "step": 5170
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001689609396999075,
+      "loss": 4.778,
+      "step": 5175
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00016892071282030653,
+      "loss": 3.5281,
+      "step": 5180
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00016888048594070558,
+      "loss": 2.7328,
+      "step": 5185
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00016884025906110464,
+      "loss": 2.6921,
+      "step": 5190
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00016880003218150367,
+      "loss": 1.5729,
+      "step": 5195
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00016875980530190273,
+      "loss": 1.321,
+      "step": 5200
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001687195784223018,
+      "loss": 3.92,
+      "step": 5205
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00016867935154270085,
+      "loss": 4.4383,
+      "step": 5210
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001686391246630999,
+      "loss": 5.2021,
+      "step": 5215
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00016859889778349894,
+      "loss": 2.773,
+      "step": 5220
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000168558670903898,
+      "loss": 3.292,
+      "step": 5225
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00016851844402429703,
+      "loss": 4.218,
+      "step": 5230
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00016847821714469609,
+      "loss": 3.38,
+      "step": 5235
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00016843799026509514,
+      "loss": 2.0585,
+      "step": 5240
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001683977633854942,
+      "loss": 3.6215,
+      "step": 5245
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00016835753650589326,
+      "loss": 0.1181,
+      "step": 5250
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001683173096262923,
+      "loss": 4.9352,
+      "step": 5255
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00016827708274669135,
+      "loss": 4.5021,
+      "step": 5260
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00016823685586709038,
+      "loss": 4.435,
+      "step": 5265
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00016819662898748944,
+      "loss": 3.6467,
+      "step": 5270
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001681564021078885,
+      "loss": 3.5488,
+      "step": 5275
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00016811617522828756,
+      "loss": 3.414,
+      "step": 5280
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00016807594834868661,
+      "loss": 3.1631,
+      "step": 5285
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00016803572146908567,
+      "loss": 2.0229,
+      "step": 5290
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001679954945894847,
+      "loss": 1.2201,
+      "step": 5295
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00016795526770988373,
+      "loss": 3.7302,
+      "step": 5300
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001679150408302828,
+      "loss": 4.5115,
+      "step": 5305
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00016787481395068185,
+      "loss": 4.2893,
+      "step": 5310
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001678345870710809,
+      "loss": 3.5971,
+      "step": 5315
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00016779436019147997,
+      "loss": 4.424,
+      "step": 5320
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00016775413331187903,
+      "loss": 3.4488,
+      "step": 5325
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00016771390643227806,
+      "loss": 2.7201,
+      "step": 5330
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001676736795526771,
+      "loss": 2.8584,
+      "step": 5335
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00016763345267307615,
+      "loss": 2.0608,
+      "step": 5340
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001675932257934752,
+      "loss": 1.7905,
+      "step": 5345
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00016755299891387426,
+      "loss": 1.7893,
+      "step": 5350
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00016751277203427332,
+      "loss": 4.6043,
+      "step": 5355
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00016747254515467238,
+      "loss": 4.7084,
+      "step": 5360
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001674323182750714,
+      "loss": 4.0279,
+      "step": 5365
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00016739209139547044,
+      "loss": 4.6014,
+      "step": 5370
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001673518645158695,
+      "loss": 4.233,
+      "step": 5375
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00016731163763626856,
+      "loss": 4.4205,
+      "step": 5380
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00016727141075666762,
+      "loss": 3.7451,
+      "step": 5385
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00016723118387706667,
+      "loss": 1.9283,
+      "step": 5390
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00016719095699746573,
+      "loss": 1.5219,
+      "step": 5395
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00016715073011786476,
+      "loss": 1.0435,
+      "step": 5400
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00016711050323826382,
+      "loss": 3.7072,
+      "step": 5405
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00016707027635866285,
+      "loss": 4.7615,
+      "step": 5410
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001670300494790619,
+      "loss": 3.8532,
+      "step": 5415
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00016698982259946097,
+      "loss": 4.3826,
+      "step": 5420
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00016694959571986003,
+      "loss": 3.766,
+      "step": 5425
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00016690936884025906,
+      "loss": 2.053,
+      "step": 5430
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00016686914196065812,
+      "loss": 3.3234,
+      "step": 5435
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00016682891508105718,
+      "loss": 1.802,
+      "step": 5440
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001667886882014562,
+      "loss": 2.0026,
+      "step": 5445
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00016674846132185527,
+      "loss": 1.087,
+      "step": 5450
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00016670823444225432,
+      "loss": 4.4006,
+      "step": 5455
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00016666800756265338,
+      "loss": 4.633,
+      "step": 5460
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001666277806830524,
+      "loss": 3.4943,
+      "step": 5465
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00016658755380345147,
+      "loss": 4.6764,
+      "step": 5470
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00016654732692385053,
+      "loss": 3.6434,
+      "step": 5475
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001665071000442496,
+      "loss": 3.2263,
+      "step": 5480
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00016646687316464862,
+      "loss": 2.8007,
+      "step": 5485
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00016642664628504768,
+      "loss": 2.9424,
+      "step": 5490
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00016638641940544674,
+      "loss": 1.1535,
+      "step": 5495
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00016634619252584577,
+      "loss": 0.2137,
+      "step": 5500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00016630596564624482,
+      "loss": 4.2225,
+      "step": 5505
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00016626573876664388,
+      "loss": 4.5811,
+      "step": 5510
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00016622551188704294,
+      "loss": 5.7295,
+      "step": 5515
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00016618528500744197,
+      "loss": 4.3728,
+      "step": 5520
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00016614505812784103,
+      "loss": 3.9244,
+      "step": 5525
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001661048312482401,
+      "loss": 3.3557,
+      "step": 5530
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00016606460436863912,
+      "loss": 2.9875,
+      "step": 5535
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00016602437748903818,
+      "loss": 3.5087,
+      "step": 5540
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00016598415060943724,
+      "loss": 3.3396,
+      "step": 5545
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001659439237298363,
+      "loss": 1.8395,
+      "step": 5550
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00016590369685023535,
+      "loss": 4.0984,
+      "step": 5555
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00016586346997063438,
+      "loss": 3.4518,
+      "step": 5560
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00016582324309103344,
+      "loss": 4.3408,
+      "step": 5565
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00016578301621143247,
+      "loss": 3.5854,
+      "step": 5570
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00016574278933183153,
+      "loss": 3.4309,
+      "step": 5575
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001657025624522306,
+      "loss": 3.1176,
+      "step": 5580
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00016566233557262965,
+      "loss": 3.166,
+      "step": 5585
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001656221086930287,
+      "loss": 2.8318,
+      "step": 5590
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00016558188181342774,
+      "loss": 1.8063,
+      "step": 5595
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001655416549338268,
+      "loss": 0.0932,
+      "step": 5600
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00016550142805422583,
+      "loss": 4.1902,
+      "step": 5605
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00016546120117462489,
+      "loss": 4.9719,
+      "step": 5610
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00016542097429502394,
+      "loss": 4.8271,
+      "step": 5615
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000165380747415423,
+      "loss": 4.3954,
+      "step": 5620
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00016534052053582206,
+      "loss": 3.5303,
+      "step": 5625
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00016530029365622112,
+      "loss": 2.2174,
+      "step": 5630
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00016526006677662015,
+      "loss": 2.1981,
+      "step": 5635
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00016521983989701918,
+      "loss": 2.1,
+      "step": 5640
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00016517961301741824,
+      "loss": 2.3089,
+      "step": 5645
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001651393861378173,
+      "loss": 3.9498,
+      "step": 5650
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00016509915925821636,
+      "loss": 4.8281,
+      "step": 5655
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001650589323786154,
+      "loss": 5.3375,
+      "step": 5660
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00016501870549901447,
+      "loss": 4.2732,
+      "step": 5665
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001649784786194135,
+      "loss": 3.8743,
+      "step": 5670
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00016493825173981253,
+      "loss": 2.8098,
+      "step": 5675
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001648980248602116,
+      "loss": 3.626,
+      "step": 5680
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00016485779798061065,
+      "loss": 3.1708,
+      "step": 5685
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001648175711010097,
+      "loss": 2.9451,
+      "step": 5690
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00016477734422140877,
+      "loss": 2.7528,
+      "step": 5695
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00016473711734180783,
+      "loss": 0.7335,
+      "step": 5700
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00016469689046220686,
+      "loss": 4.0922,
+      "step": 5705
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001646566635826059,
+      "loss": 4.3029,
+      "step": 5710
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00016461643670300495,
+      "loss": 4.0666,
+      "step": 5715
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000164576209823404,
+      "loss": 4.3164,
+      "step": 5720
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00016453598294380306,
+      "loss": 3.5441,
+      "step": 5725
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00016449575606420212,
+      "loss": 2.875,
+      "step": 5730
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00016445552918460115,
+      "loss": 2.0636,
+      "step": 5735
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001644153023050002,
+      "loss": 2.9502,
+      "step": 5740
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00016437507542539924,
+      "loss": 0.608,
+      "step": 5745
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001643348485457983,
+      "loss": 2.9006,
+      "step": 5750
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00016429462166619736,
+      "loss": 5.0873,
+      "step": 5755
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00016425439478659642,
+      "loss": 4.1059,
+      "step": 5760
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00016421416790699547,
+      "loss": 3.5555,
+      "step": 5765
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001641739410273945,
+      "loss": 3.1217,
+      "step": 5770
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00016413371414779356,
+      "loss": 3.9523,
+      "step": 5775
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00016409348726819262,
+      "loss": 2.5499,
+      "step": 5780
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00016405326038859165,
+      "loss": 1.9861,
+      "step": 5785
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001640130335089907,
+      "loss": 1.938,
+      "step": 5790
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00016397280662938977,
+      "loss": 1.8224,
+      "step": 5795
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00016393257974978883,
+      "loss": 3.7677,
+      "step": 5800
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00016389235287018786,
+      "loss": 4.5553,
+      "step": 5805
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00016385212599058692,
+      "loss": 4.0107,
+      "step": 5810
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00016381189911098597,
+      "loss": 4.0619,
+      "step": 5815
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000163771672231385,
+      "loss": 4.0109,
+      "step": 5820
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00016373144535178406,
+      "loss": 3.6002,
+      "step": 5825
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00016369121847218312,
+      "loss": 3.1553,
+      "step": 5830
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00016365099159258218,
+      "loss": 2.579,
+      "step": 5835
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001636107647129812,
+      "loss": 2.6138,
+      "step": 5840
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00016357053783338027,
+      "loss": 1.2922,
+      "step": 5845
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00016353031095377933,
+      "loss": 1.3788,
+      "step": 5850
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001634900840741784,
+      "loss": 3.92,
+      "step": 5855
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00016344985719457742,
+      "loss": 4.1422,
+      "step": 5860
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00016340963031497648,
+      "loss": 4.1127,
+      "step": 5865
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00016336940343537553,
+      "loss": 3.7213,
+      "step": 5870
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00016332917655577457,
+      "loss": 3.4097,
+      "step": 5875
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00016328894967617362,
+      "loss": 2.2418,
+      "step": 5880
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00016324872279657268,
+      "loss": 2.4848,
+      "step": 5885
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00016320849591697174,
+      "loss": 2.2139,
+      "step": 5890
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00016316826903737077,
+      "loss": 2.1187,
+      "step": 5895
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00016312804215776983,
+      "loss": 0.5452,
+      "step": 5900
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001630878152781689,
+      "loss": 5.1633,
+      "step": 5905
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00016304758839856792,
+      "loss": 5.2154,
+      "step": 5910
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00016300736151896698,
+      "loss": 4.3537,
+      "step": 5915
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00016296713463936604,
+      "loss": 4.3426,
+      "step": 5920
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001629269077597651,
+      "loss": 3.8029,
+      "step": 5925
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00016288668088016415,
+      "loss": 3.3158,
+      "step": 5930
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00016284645400056318,
+      "loss": 2.3386,
+      "step": 5935
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00016280622712096224,
+      "loss": 2.2401,
+      "step": 5940
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00016276600024136127,
+      "loss": 3.1957,
+      "step": 5945
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00016272577336176033,
+      "loss": 3.0096,
+      "step": 5950
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001626855464821594,
+      "loss": 4.1182,
+      "step": 5955
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00016264531960255845,
+      "loss": 4.3215,
+      "step": 5960
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001626050927229575,
+      "loss": 4.4809,
+      "step": 5965
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00016256486584335654,
+      "loss": 4.4412,
+      "step": 5970
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001625246389637556,
+      "loss": 5.4105,
+      "step": 5975
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00016248441208415463,
+      "loss": 4.3814,
+      "step": 5980
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00016244418520455368,
+      "loss": 2.7278,
+      "step": 5985
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00016240395832495274,
+      "loss": 3.1769,
+      "step": 5990
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001623637314453518,
+      "loss": 0.9617,
+      "step": 5995
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00016232350456575086,
+      "loss": 1.1335,
+      "step": 6000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00016228327768614992,
+      "loss": 4.5203,
+      "step": 6005
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00016224305080654895,
+      "loss": 4.9381,
+      "step": 6010
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00016220282392694798,
+      "loss": 4.7555,
+      "step": 6015
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00016216259704734704,
+      "loss": 4.7895,
+      "step": 6020
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001621223701677461,
+      "loss": 3.355,
+      "step": 6025
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00016208214328814515,
+      "loss": 3.2165,
+      "step": 6030
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001620419164085442,
+      "loss": 3.9241,
+      "step": 6035
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00016200168952894324,
+      "loss": 2.3204,
+      "step": 6040
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00016196146264934227,
+      "loss": 1.1676,
+      "step": 6045
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00016192123576974133,
+      "loss": 0.6222,
+      "step": 6050
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001618810088901404,
+      "loss": 4.2146,
+      "step": 6055
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00016184078201053945,
+      "loss": 5.6813,
+      "step": 6060
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001618005551309385,
+      "loss": 3.683,
+      "step": 6065
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00016176032825133757,
+      "loss": 3.5365,
+      "step": 6070
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001617201013717366,
+      "loss": 2.232,
+      "step": 6075
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00016167987449213566,
+      "loss": 3.2946,
+      "step": 6080
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00016163964761253469,
+      "loss": 2.1471,
+      "step": 6085
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00016159942073293374,
+      "loss": 3.233,
+      "step": 6090
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001615591938533328,
+      "loss": 3.6617,
+      "step": 6095
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00016151896697373186,
+      "loss": 1.3713,
+      "step": 6100
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00016147874009413092,
+      "loss": 5.2553,
+      "step": 6105
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00016143851321452995,
+      "loss": 4.0039,
+      "step": 6110
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000161398286334929,
+      "loss": 4.5361,
+      "step": 6115
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00016135805945532804,
+      "loss": 3.3616,
+      "step": 6120
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001613178325757271,
+      "loss": 3.3566,
+      "step": 6125
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00016127760569612616,
+      "loss": 3.7517,
+      "step": 6130
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00016123737881652521,
+      "loss": 3.2124,
+      "step": 6135
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00016119715193692427,
+      "loss": 2.3577,
+      "step": 6140
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001611569250573233,
+      "loss": 3.5035,
+      "step": 6145
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00016111669817772236,
+      "loss": 1.3117,
+      "step": 6150
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00016107647129812142,
+      "loss": 4.2394,
+      "step": 6155
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00016103624441852045,
+      "loss": 4.5457,
+      "step": 6160
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001609960175389195,
+      "loss": 3.7481,
+      "step": 6165
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00016095579065931857,
+      "loss": 3.9391,
+      "step": 6170
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00016091556377971763,
+      "loss": 3.5035,
+      "step": 6175
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00016087533690011666,
+      "loss": 3.3803,
+      "step": 6180
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00016083511002051572,
+      "loss": 2.449,
+      "step": 6185
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00016079488314091477,
+      "loss": 1.8994,
+      "step": 6190
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001607546562613138,
+      "loss": 2.9759,
+      "step": 6195
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00016071442938171286,
+      "loss": 2.7731,
+      "step": 6200
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00016067420250211192,
+      "loss": 4.325,
+      "step": 6205
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00016063397562251098,
+      "loss": 4.6818,
+      "step": 6210
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00016059374874291,
+      "loss": 3.3004,
+      "step": 6215
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00016055352186330907,
+      "loss": 3.7553,
+      "step": 6220
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00016051329498370813,
+      "loss": 3.8145,
+      "step": 6225
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00016047306810410719,
+      "loss": 3.1997,
+      "step": 6230
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00016043284122450622,
+      "loss": 3.2375,
+      "step": 6235
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00016039261434490528,
+      "loss": 2.0859,
+      "step": 6240
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00016035238746530433,
+      "loss": 3.2224,
+      "step": 6245
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00016031216058570336,
+      "loss": 2.5072,
+      "step": 6250
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00016027193370610242,
+      "loss": 4.9318,
+      "step": 6255
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00016023170682650148,
+      "loss": 4.1885,
+      "step": 6260
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00016019147994690054,
+      "loss": 4.3771,
+      "step": 6265
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00016015125306729957,
+      "loss": 4.267,
+      "step": 6270
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00016011102618769863,
+      "loss": 4.1869,
+      "step": 6275
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001600707993080977,
+      "loss": 2.5321,
+      "step": 6280
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00016003057242849672,
+      "loss": 2.5823,
+      "step": 6285
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00015999034554889578,
+      "loss": 1.3439,
+      "step": 6290
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00015995011866929483,
+      "loss": 2.2149,
+      "step": 6295
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001599098917896939,
+      "loss": 1.6957,
+      "step": 6300
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00015986966491009295,
+      "loss": 4.6896,
+      "step": 6305
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00015982943803049198,
+      "loss": 4.6721,
+      "step": 6310
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00015978921115089104,
+      "loss": 3.8354,
+      "step": 6315
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00015974898427129007,
+      "loss": 3.1878,
+      "step": 6320
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00015970875739168913,
+      "loss": 3.6049,
+      "step": 6325
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001596685305120882,
+      "loss": 3.4877,
+      "step": 6330
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00015962830363248725,
+      "loss": 3.7444,
+      "step": 6335
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001595880767528863,
+      "loss": 2.6608,
+      "step": 6340
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00015954784987328534,
+      "loss": 1.8235,
+      "step": 6345
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00015950762299368437,
+      "loss": 1.9221,
+      "step": 6350
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00015946739611408342,
+      "loss": 4.5215,
+      "step": 6355
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00015942716923448248,
+      "loss": 5.2854,
+      "step": 6360
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00015938694235488154,
+      "loss": 4.3033,
+      "step": 6365
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001593467154752806,
+      "loss": 5.4211,
+      "step": 6370
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00015930648859567966,
+      "loss": 2.9237,
+      "step": 6375
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001592662617160787,
+      "loss": 3.6648,
+      "step": 6380
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00015922603483647772,
+      "loss": 2.2996,
+      "step": 6385
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00015918580795687678,
+      "loss": 2.3559,
+      "step": 6390
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00015914558107727584,
+      "loss": 1.8411,
+      "step": 6395
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001591053541976749,
+      "loss": 1.8784,
+      "step": 6400
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00015906512731807395,
+      "loss": 3.5871,
+      "step": 6405
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000159024900438473,
+      "loss": 4.8471,
+      "step": 6410
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00015898467355887204,
+      "loss": 3.7787,
+      "step": 6415
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00015894444667927107,
+      "loss": 3.7279,
+      "step": 6420
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00015890421979967013,
+      "loss": 4.0128,
+      "step": 6425
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001588639929200692,
+      "loss": 3.1311,
+      "step": 6430
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00015882376604046825,
+      "loss": 2.5843,
+      "step": 6435
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001587835391608673,
+      "loss": 4.284,
+      "step": 6440
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00015874331228126636,
+      "loss": 3.2978,
+      "step": 6445
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001587030854016654,
+      "loss": 0.0883,
+      "step": 6450
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00015866285852206445,
+      "loss": 3.7918,
+      "step": 6455
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00015862263164246349,
+      "loss": 4.1014,
+      "step": 6460
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00015858240476286254,
+      "loss": 3.642,
+      "step": 6465
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001585421778832616,
+      "loss": 4.5049,
+      "step": 6470
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00015850195100366066,
+      "loss": 3.2287,
+      "step": 6475
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00015846172412405972,
+      "loss": 2.9815,
+      "step": 6480
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00015842149724445875,
+      "loss": 3.578,
+      "step": 6485
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001583812703648578,
+      "loss": 3.4713,
+      "step": 6490
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00015834104348525684,
+      "loss": 2.6449,
+      "step": 6495
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001583008166056559,
+      "loss": 3.3023,
+      "step": 6500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00015826058972605496,
+      "loss": 3.7168,
+      "step": 6505
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00015822036284645401,
+      "loss": 3.6531,
+      "step": 6510
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00015818013596685307,
+      "loss": 4.1709,
+      "step": 6515
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001581399090872521,
+      "loss": 3.7805,
+      "step": 6520
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00015809968220765116,
+      "loss": 4.2029,
+      "step": 6525
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00015805945532805022,
+      "loss": 3.6784,
+      "step": 6530
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00015801922844844925,
+      "loss": 2.9541,
+      "step": 6535
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001579790015688483,
+      "loss": 2.3555,
+      "step": 6540
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00015793877468924737,
+      "loss": 2.8182,
+      "step": 6545
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00015789854780964643,
+      "loss": 0.6913,
+      "step": 6550
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00015785832093004546,
+      "loss": 3.7799,
+      "step": 6555
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00015781809405044451,
+      "loss": 4.8266,
+      "step": 6560
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00015777786717084357,
+      "loss": 4.3416,
+      "step": 6565
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001577376402912426,
+      "loss": 3.3998,
+      "step": 6570
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00015769741341164166,
+      "loss": 3.0765,
+      "step": 6575
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00015765718653204072,
+      "loss": 3.1966,
+      "step": 6580
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00015761695965243978,
+      "loss": 3.4283,
+      "step": 6585
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001575767327728388,
+      "loss": 1.7517,
+      "step": 6590
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00015753650589323787,
+      "loss": 2.4111,
+      "step": 6595
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00015749627901363693,
+      "loss": 1.31,
+      "step": 6600
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00015745605213403598,
+      "loss": 5.1953,
+      "step": 6605
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00015741582525443502,
+      "loss": 4.7135,
+      "step": 6610
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00015737559837483407,
+      "loss": 3.4439,
+      "step": 6615
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00015733537149523313,
+      "loss": 4.5059,
+      "step": 6620
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00015729514461563216,
+      "loss": 4.826,
+      "step": 6625
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00015725491773603122,
+      "loss": 4.1916,
+      "step": 6630
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00015721469085643028,
+      "loss": 3.258,
+      "step": 6635
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00015717446397682934,
+      "loss": 2.7372,
+      "step": 6640
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00015713423709722837,
+      "loss": 2.4825,
+      "step": 6645
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00015709401021762743,
+      "loss": 0.8767,
+      "step": 6650
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00015705378333802646,
+      "loss": 5.0094,
+      "step": 6655
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00015701355645842552,
+      "loss": 5.1865,
+      "step": 6660
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00015697332957882458,
+      "loss": 3.6629,
+      "step": 6665
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00015693310269922363,
+      "loss": 4.3094,
+      "step": 6670
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001568928758196227,
+      "loss": 3.6267,
+      "step": 6675
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00015685264894002175,
+      "loss": 3.056,
+      "step": 6680
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00015681242206042078,
+      "loss": 2.7199,
+      "step": 6685
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001567721951808198,
+      "loss": 2.7694,
+      "step": 6690
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00015673196830121887,
+      "loss": 3.5551,
+      "step": 6695
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00015669174142161793,
+      "loss": 0.0749,
+      "step": 6700
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000156651514542017,
+      "loss": 4.768,
+      "step": 6705
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00015661128766241605,
+      "loss": 4.2012,
+      "step": 6710
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001565710607828151,
+      "loss": 2.9939,
+      "step": 6715
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00015653083390321413,
+      "loss": 2.8134,
+      "step": 6720
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00015649060702361317,
+      "loss": 3.4247,
+      "step": 6725
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00015645038014401222,
+      "loss": 3.0152,
+      "step": 6730
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00015641015326441128,
+      "loss": 2.0896,
+      "step": 6735
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00015636992638481034,
+      "loss": 2.5528,
+      "step": 6740
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001563296995052094,
+      "loss": 0.7059,
+      "step": 6745
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00015628947262560846,
+      "loss": 1.9621,
+      "step": 6750
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001562492457460075,
+      "loss": 5.2277,
+      "step": 6755
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00015620901886640652,
+      "loss": 5.1182,
+      "step": 6760
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00015616879198680558,
+      "loss": 4.7277,
+      "step": 6765
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00015612856510720464,
+      "loss": 3.9717,
+      "step": 6770
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001560883382276037,
+      "loss": 3.8693,
+      "step": 6775
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00015604811134800275,
+      "loss": 2.4765,
+      "step": 6780
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001560078844684018,
+      "loss": 3.0296,
+      "step": 6785
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00015596765758880084,
+      "loss": 2.2676,
+      "step": 6790
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00015592743070919987,
+      "loss": 2.3231,
+      "step": 6795
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00015588720382959893,
+      "loss": 0.0427,
+      "step": 6800
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000155846976949998,
+      "loss": 4.8117,
+      "step": 6805
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00015580675007039705,
+      "loss": 4.2992,
+      "step": 6810
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001557665231907961,
+      "loss": 4.5337,
+      "step": 6815
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00015572629631119516,
+      "loss": 5.334,
+      "step": 6820
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001556860694315942,
+      "loss": 3.5176,
+      "step": 6825
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00015564584255199325,
+      "loss": 1.5627,
+      "step": 6830
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00015560561567239228,
+      "loss": 2.1864,
+      "step": 6835
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00015556538879279134,
+      "loss": 0.7217,
+      "step": 6840
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001555251619131904,
+      "loss": 3.6855,
+      "step": 6845
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00015548493503358946,
+      "loss": 1.8674,
+      "step": 6850
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00015544470815398852,
+      "loss": 4.4283,
+      "step": 6855
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00015540448127438755,
+      "loss": 4.7135,
+      "step": 6860
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001553642543947866,
+      "loss": 3.5176,
+      "step": 6865
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00015532402751518564,
+      "loss": 5.2475,
+      "step": 6870
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001552838006355847,
+      "loss": 3.5299,
+      "step": 6875
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00015524357375598375,
+      "loss": 3.6828,
+      "step": 6880
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001552033468763828,
+      "loss": 5.0184,
+      "step": 6885
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00015516311999678187,
+      "loss": 3.1216,
+      "step": 6890
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001551228931171809,
+      "loss": 3.048,
+      "step": 6895
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00015508266623757996,
+      "loss": 3.0663,
+      "step": 6900
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00015504243935797902,
+      "loss": 5.0727,
+      "step": 6905
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00015500221247837805,
+      "loss": 4.3605,
+      "step": 6910
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001549619855987771,
+      "loss": 4.4473,
+      "step": 6915
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00015492175871917617,
+      "loss": 3.3799,
+      "step": 6920
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00015488153183957522,
+      "loss": 3.7008,
+      "step": 6925
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00015484130495997426,
+      "loss": 3.7081,
+      "step": 6930
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00015480107808037331,
+      "loss": 3.5293,
+      "step": 6935
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00015476085120077237,
+      "loss": 2.8219,
+      "step": 6940
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001547206243211714,
+      "loss": 2.5409,
+      "step": 6945
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00015468039744157046,
+      "loss": 1.8382,
+      "step": 6950
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00015464017056196952,
+      "loss": 4.9234,
+      "step": 6955
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00015459994368236855,
+      "loss": 5.3937,
+      "step": 6960
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001545597168027676,
+      "loss": 4.5607,
+      "step": 6965
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00015451948992316667,
+      "loss": 3.7303,
+      "step": 6970
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00015447926304356573,
+      "loss": 3.9004,
+      "step": 6975
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00015443903616396478,
+      "loss": 3.7896,
+      "step": 6980
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00015439880928436381,
+      "loss": 2.2128,
+      "step": 6985
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00015435858240476287,
+      "loss": 1.8755,
+      "step": 6990
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001543183555251619,
+      "loss": 2.1864,
+      "step": 6995
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00015427812864556096,
+      "loss": 0.8963,
+      "step": 7000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00015423790176596002,
+      "loss": 4.0704,
+      "step": 7005
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00015419767488635908,
+      "loss": 4.6586,
+      "step": 7010
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00015415744800675814,
+      "loss": 4.1082,
+      "step": 7015
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00015411722112715717,
+      "loss": 3.795,
+      "step": 7020
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00015407699424755623,
+      "loss": 4.5971,
+      "step": 7025
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00015403676736795526,
+      "loss": 4.1455,
+      "step": 7030
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00015399654048835432,
+      "loss": 2.7245,
+      "step": 7035
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00015395631360875337,
+      "loss": 3.3336,
+      "step": 7040
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00015391608672915243,
+      "loss": 3.3736,
+      "step": 7045
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001538758598495515,
+      "loss": 2.5276,
+      "step": 7050
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00015383563296995055,
+      "loss": 5.3309,
+      "step": 7055
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00015379540609034958,
+      "loss": 4.858,
+      "step": 7060
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001537551792107486,
+      "loss": 4.5324,
+      "step": 7065
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00015371495233114767,
+      "loss": 4.2461,
+      "step": 7070
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00015367472545154673,
+      "loss": 3.8908,
+      "step": 7075
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00015363449857194579,
+      "loss": 3.1788,
+      "step": 7080
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00015359427169234484,
+      "loss": 3.3179,
+      "step": 7085
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001535540448127439,
+      "loss": 2.7894,
+      "step": 7090
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00015351381793314293,
+      "loss": 2.9542,
+      "step": 7095
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00015347359105354196,
+      "loss": 0.7846,
+      "step": 7100
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00015343336417394102,
+      "loss": 5.3049,
+      "step": 7105
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00015339313729434008,
+      "loss": 4.2867,
+      "step": 7110
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00015335291041473914,
+      "loss": 4.4562,
+      "step": 7115
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001533126835351382,
+      "loss": 3.9077,
+      "step": 7120
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00015327245665553726,
+      "loss": 3.4854,
+      "step": 7125
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001532322297759363,
+      "loss": 2.647,
+      "step": 7130
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00015319200289633532,
+      "loss": 2.2412,
+      "step": 7135
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00015315177601673438,
+      "loss": 2.5988,
+      "step": 7140
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00015311154913713343,
+      "loss": 2.807,
+      "step": 7145
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001530713222575325,
+      "loss": 3.1684,
+      "step": 7150
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00015303109537793155,
+      "loss": 4.0076,
+      "step": 7155
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001529908684983306,
+      "loss": 3.8545,
+      "step": 7160
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00015295064161872964,
+      "loss": 4.7838,
+      "step": 7165
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00015291041473912867,
+      "loss": 4.7896,
+      "step": 7170
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00015287018785952773,
+      "loss": 4.5879,
+      "step": 7175
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001528299609799268,
+      "loss": 4.1828,
+      "step": 7180
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00015278973410032585,
+      "loss": 3.6771,
+      "step": 7185
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001527495072207249,
+      "loss": 2.4002,
+      "step": 7190
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00015270928034112396,
+      "loss": 2.9123,
+      "step": 7195
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000152669053461523,
+      "loss": 1.4825,
+      "step": 7200
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00015262882658192205,
+      "loss": 4.3521,
+      "step": 7205
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00015258859970232108,
+      "loss": 4.76,
+      "step": 7210
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00015254837282272014,
+      "loss": 4.7199,
+      "step": 7215
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001525081459431192,
+      "loss": 3.7285,
+      "step": 7220
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00015246791906351826,
+      "loss": 4.2059,
+      "step": 7225
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00015242769218391732,
+      "loss": 2.4594,
+      "step": 7230
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00015238746530431635,
+      "loss": 1.7533,
+      "step": 7235
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001523472384247154,
+      "loss": 2.0827,
+      "step": 7240
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00015230701154511444,
+      "loss": 3.332,
+      "step": 7245
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001522667846655135,
+      "loss": 1.1347,
+      "step": 7250
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00015222655778591255,
+      "loss": 3.9932,
+      "step": 7255
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001521863309063116,
+      "loss": 3.5039,
+      "step": 7260
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00015214610402671064,
+      "loss": 3.6082,
+      "step": 7265
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001521058771471097,
+      "loss": 2.8597,
+      "step": 7270
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00015206565026750876,
+      "loss": 3.6561,
+      "step": 7275
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00015202542338790782,
+      "loss": 1.5933,
+      "step": 7280
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00015198519650830685,
+      "loss": 2.386,
+      "step": 7285
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001519449696287059,
+      "loss": 2.4073,
+      "step": 7290
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00015190474274910497,
+      "loss": 2.3714,
+      "step": 7295
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000151864515869504,
+      "loss": 1.211,
+      "step": 7300
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00015182428898990305,
+      "loss": 3.9281,
+      "step": 7305
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001517840621103021,
+      "loss": 4.6617,
+      "step": 7310
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00015174383523070117,
+      "loss": 4.624,
+      "step": 7315
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001517036083511002,
+      "loss": 4.2254,
+      "step": 7320
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00015166338147149926,
+      "loss": 3.4564,
+      "step": 7325
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00015162315459189832,
+      "loss": 3.0671,
+      "step": 7330
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00015158292771229735,
+      "loss": 2.1117,
+      "step": 7335
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001515427008326964,
+      "loss": 2.8584,
+      "step": 7340
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00015150247395309547,
+      "loss": 2.352,
+      "step": 7345
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00015146224707349452,
+      "loss": 1.9422,
+      "step": 7350
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00015142202019389358,
+      "loss": 4.873,
+      "step": 7355
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00015138179331429261,
+      "loss": 4.7299,
+      "step": 7360
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00015134156643469167,
+      "loss": 3.7564,
+      "step": 7365
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001513013395550907,
+      "loss": 5.5928,
+      "step": 7370
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00015126111267548976,
+      "loss": 4.1801,
+      "step": 7375
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00015122088579588882,
+      "loss": 3.195,
+      "step": 7380
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00015118065891628788,
+      "loss": 2.6585,
+      "step": 7385
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00015114043203668694,
+      "loss": 2.2449,
+      "step": 7390
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00015110020515708597,
+      "loss": 3.2964,
+      "step": 7395
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00015105997827748503,
+      "loss": 1.8592,
+      "step": 7400
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00015101975139788406,
+      "loss": 5.1701,
+      "step": 7405
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00015097952451828312,
+      "loss": 3.9732,
+      "step": 7410
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00015093929763868217,
+      "loss": 3.7973,
+      "step": 7415
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00015089907075908123,
+      "loss": 4.9865,
+      "step": 7420
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001508588438794803,
+      "loss": 3.4799,
+      "step": 7425
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00015081861699987935,
+      "loss": 2.9448,
+      "step": 7430
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00015077839012027838,
+      "loss": 2.2613,
+      "step": 7435
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001507381632406774,
+      "loss": 3.6979,
+      "step": 7440
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00015069793636107647,
+      "loss": 1.1705,
+      "step": 7445
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00015065770948147553,
+      "loss": 1.0137,
+      "step": 7450
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00015061748260187459,
+      "loss": 4.4939,
+      "step": 7455
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00015057725572227364,
+      "loss": 4.7318,
+      "step": 7460
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001505370288426727,
+      "loss": 4.4439,
+      "step": 7465
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00015049680196307173,
+      "loss": 3.9766,
+      "step": 7470
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00015045657508347076,
+      "loss": 4.8434,
+      "step": 7475
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00015041634820386982,
+      "loss": 3.7546,
+      "step": 7480
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00015037612132426888,
+      "loss": 4.257,
+      "step": 7485
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00015033589444466794,
+      "loss": 2.0724,
+      "step": 7490
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000150295667565067,
+      "loss": 2.3552,
+      "step": 7495
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00015025544068546606,
+      "loss": 2.8831,
+      "step": 7500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00015021521380586509,
+      "loss": 4.2111,
+      "step": 7505
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00015017498692626412,
+      "loss": 3.5881,
+      "step": 7510
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00015013476004666318,
+      "loss": 3.7628,
+      "step": 7515
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00015009453316706223,
+      "loss": 4.3022,
+      "step": 7520
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001500543062874613,
+      "loss": 3.9914,
+      "step": 7525
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015001407940786035,
+      "loss": 3.0919,
+      "step": 7530
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001499738525282594,
+      "loss": 3.287,
+      "step": 7535
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00014993362564865844,
+      "loss": 3.2431,
+      "step": 7540
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00014989339876905747,
+      "loss": 1.7876,
+      "step": 7545
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00014985317188945653,
+      "loss": 3.392,
+      "step": 7550
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001498129450098556,
+      "loss": 4.3033,
+      "step": 7555
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00014977271813025465,
+      "loss": 4.7418,
+      "step": 7560
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001497324912506537,
+      "loss": 3.6264,
+      "step": 7565
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00014969226437105273,
+      "loss": 3.9139,
+      "step": 7570
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001496520374914518,
+      "loss": 4.6238,
+      "step": 7575
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00014961181061185085,
+      "loss": 3.354,
+      "step": 7580
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00014957158373224988,
+      "loss": 2.212,
+      "step": 7585
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00014953135685264894,
+      "loss": 2.9092,
+      "step": 7590
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000149491129973048,
+      "loss": 1.8592,
+      "step": 7595
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00014945090309344706,
+      "loss": 0.1596,
+      "step": 7600
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001494106762138461,
+      "loss": 3.8275,
+      "step": 7605
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00014937044933424515,
+      "loss": 3.526,
+      "step": 7610
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001493302224546442,
+      "loss": 4.7551,
+      "step": 7615
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00014928999557504324,
+      "loss": 2.7891,
+      "step": 7620
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001492497686954423,
+      "loss": 3.836,
+      "step": 7625
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00014920954181584135,
+      "loss": 4.1405,
+      "step": 7630
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001491693149362404,
+      "loss": 3.3979,
+      "step": 7635
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00014912908805663944,
+      "loss": 2.051,
+      "step": 7640
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001490888611770385,
+      "loss": 2.0498,
+      "step": 7645
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00014904863429743756,
+      "loss": 1.2102,
+      "step": 7650
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00014900840741783662,
+      "loss": 4.4957,
+      "step": 7655
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00014896818053823565,
+      "loss": 4.0664,
+      "step": 7660
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001489279536586347,
+      "loss": 4.7891,
+      "step": 7665
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00014888772677903376,
+      "loss": 2.6803,
+      "step": 7670
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001488474998994328,
+      "loss": 3.4743,
+      "step": 7675
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00014880727301983185,
+      "loss": 3.2174,
+      "step": 7680
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001487670461402309,
+      "loss": 3.109,
+      "step": 7685
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00014872681926062997,
+      "loss": 1.0869,
+      "step": 7690
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00014868659238102903,
+      "loss": 1.1583,
+      "step": 7695
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00014864636550142806,
+      "loss": 3.8975,
+      "step": 7700
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00014860613862182712,
+      "loss": 5.1785,
+      "step": 7705
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00014856591174222615,
+      "loss": 3.8462,
+      "step": 7710
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001485256848626252,
+      "loss": 3.6188,
+      "step": 7715
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00014848545798302427,
+      "loss": 3.4301,
+      "step": 7720
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00014844523110342332,
+      "loss": 4.3648,
+      "step": 7725
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00014840500422382238,
+      "loss": 1.6885,
+      "step": 7730
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001483647773442214,
+      "loss": 2.2321,
+      "step": 7735
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00014832455046462047,
+      "loss": 2.4661,
+      "step": 7740
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001482843235850195,
+      "loss": 3.3984,
+      "step": 7745
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00014824409670541856,
+      "loss": 1.2363,
+      "step": 7750
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00014820386982581762,
+      "loss": 5.4285,
+      "step": 7755
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00014816364294621668,
+      "loss": 3.917,
+      "step": 7760
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00014812341606661574,
+      "loss": 4.5963,
+      "step": 7765
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001480831891870148,
+      "loss": 3.875,
+      "step": 7770
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00014804296230741382,
+      "loss": 3.8596,
+      "step": 7775
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00014800273542781286,
+      "loss": 3.2287,
+      "step": 7780
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014796250854821191,
+      "loss": 3.3803,
+      "step": 7785
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014792228166861097,
+      "loss": 2.5698,
+      "step": 7790
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014788205478901003,
+      "loss": 1.9961,
+      "step": 7795
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001478418279094091,
+      "loss": 1.7948,
+      "step": 7800
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014780160102980815,
+      "loss": 4.4584,
+      "step": 7805
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014776137415020718,
+      "loss": 4.4805,
+      "step": 7810
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001477211472706062,
+      "loss": 4.1227,
+      "step": 7815
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014768092039100527,
+      "loss": 4.8541,
+      "step": 7820
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014764069351140433,
+      "loss": 3.4299,
+      "step": 7825
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014760046663180338,
+      "loss": 4.4625,
+      "step": 7830
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014756023975220244,
+      "loss": 2.7753,
+      "step": 7835
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001475200128726015,
+      "loss": 2.1818,
+      "step": 7840
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014747978599300053,
+      "loss": 0.9003,
+      "step": 7845
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014743955911339956,
+      "loss": 2.6693,
+      "step": 7850
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014739933223379862,
+      "loss": 4.8398,
+      "step": 7855
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014735910535419768,
+      "loss": 4.6248,
+      "step": 7860
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014731887847459674,
+      "loss": 4.7467,
+      "step": 7865
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001472786515949958,
+      "loss": 4.7848,
+      "step": 7870
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014723842471539483,
+      "loss": 5.0921,
+      "step": 7875
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014719819783579389,
+      "loss": 3.9907,
+      "step": 7880
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014715797095619292,
+      "loss": 4.1201,
+      "step": 7885
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014711774407659197,
+      "loss": 1.9702,
+      "step": 7890
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014707751719699103,
+      "loss": 2.9985,
+      "step": 7895
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001470372903173901,
+      "loss": 1.4111,
+      "step": 7900
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014699706343778915,
+      "loss": 4.8258,
+      "step": 7905
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014695683655818818,
+      "loss": 4.6049,
+      "step": 7910
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014691660967858724,
+      "loss": 4.159,
+      "step": 7915
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001468763827989863,
+      "loss": 3.5428,
+      "step": 7920
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00014683615591938533,
+      "loss": 3.757,
+      "step": 7925
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00014679592903978439,
+      "loss": 3.7508,
+      "step": 7930
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00014675570216018344,
+      "loss": 3.3074,
+      "step": 7935
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001467154752805825,
+      "loss": 3.104,
+      "step": 7940
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00014667524840098153,
+      "loss": 2.4619,
+      "step": 7945
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001466350215213806,
+      "loss": 2.2569,
+      "step": 7950
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00014659479464177965,
+      "loss": 3.9715,
+      "step": 7955
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00014655456776217868,
+      "loss": 3.9365,
+      "step": 7960
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00014651434088257774,
+      "loss": 4.6123,
+      "step": 7965
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001464741140029768,
+      "loss": 4.1201,
+      "step": 7970
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00014643388712337586,
+      "loss": 2.651,
+      "step": 7975
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001463936602437749,
+      "loss": 3.2354,
+      "step": 7980
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00014635343336417395,
+      "loss": 2.7716,
+      "step": 7985
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000146313206484573,
+      "loss": 1.5226,
+      "step": 7990
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00014627297960497206,
+      "loss": 1.8045,
+      "step": 7995
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001462327527253711,
+      "loss": 2.6607,
+      "step": 8000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00014619252584577015,
+      "loss": 5.0256,
+      "step": 8005
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001461522989661692,
+      "loss": 4.3643,
+      "step": 8010
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00014611207208656824,
+      "loss": 4.3105,
+      "step": 8015
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001460718452069673,
+      "loss": 4.2754,
+      "step": 8020
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00014603161832736636,
+      "loss": 4.1685,
+      "step": 8025
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00014599139144776542,
+      "loss": 4.0061,
+      "step": 8030
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00014595116456816445,
+      "loss": 2.8045,
+      "step": 8035
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001459109376885635,
+      "loss": 3.6893,
+      "step": 8040
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00014587071080896256,
+      "loss": 3.1893,
+      "step": 8045
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001458304839293616,
+      "loss": 2.4559,
+      "step": 8050
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00014579025704976065,
+      "loss": 4.7047,
+      "step": 8055
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001457500301701597,
+      "loss": 5.4594,
+      "step": 8060
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00014570980329055877,
+      "loss": 3.6199,
+      "step": 8065
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00014566957641095783,
+      "loss": 3.1002,
+      "step": 8070
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00014562934953135686,
+      "loss": 3.0232,
+      "step": 8075
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00014558912265175592,
+      "loss": 4.4474,
+      "step": 8080
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00014554889577215495,
+      "loss": 2.7882,
+      "step": 8085
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.000145508668892554,
+      "loss": 2.3826,
+      "step": 8090
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00014546844201295306,
+      "loss": 1.9198,
+      "step": 8095
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00014542821513335212,
+      "loss": 1.3647,
+      "step": 8100
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00014538798825375118,
+      "loss": 5.1926,
+      "step": 8105
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001453477613741502,
+      "loss": 3.6435,
+      "step": 8110
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00014530753449454927,
+      "loss": 4.091,
+      "step": 8115
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001452673076149483,
+      "loss": 4.3703,
+      "step": 8120
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00014522708073534736,
+      "loss": 4.308,
+      "step": 8125
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00014518685385574642,
+      "loss": 2.2132,
+      "step": 8130
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00014514662697614548,
+      "loss": 2.4709,
+      "step": 8135
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00014510640009654453,
+      "loss": 2.6757,
+      "step": 8140
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001450661732169436,
+      "loss": 2.0093,
+      "step": 8145
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00014502594633734262,
+      "loss": 1.4906,
+      "step": 8150
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00014498571945774166,
+      "loss": 4.9254,
+      "step": 8155
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001449454925781407,
+      "loss": 4.8832,
+      "step": 8160
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00014490526569853977,
+      "loss": 4.3543,
+      "step": 8165
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00014486503881893883,
+      "loss": 4.5822,
+      "step": 8170
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001448248119393379,
+      "loss": 3.6982,
+      "step": 8175
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00014478458505973692,
+      "loss": 2.7762,
+      "step": 8180
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00014474435818013598,
+      "loss": 3.1775,
+      "step": 8185
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.000144704131300535,
+      "loss": 1.7506,
+      "step": 8190
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00014466390442093407,
+      "loss": 2.6996,
+      "step": 8195
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00014462367754133312,
+      "loss": 1.8007,
+      "step": 8200
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00014458345066173218,
+      "loss": 5.0187,
+      "step": 8205
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00014454322378213124,
+      "loss": 3.435,
+      "step": 8210
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00014450299690253027,
+      "loss": 3.5404,
+      "step": 8215
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00014446277002292933,
+      "loss": 5.1195,
+      "step": 8220
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00014442254314332836,
+      "loss": 3.7687,
+      "step": 8225
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00014438231626372742,
+      "loss": 4.91,
+      "step": 8230
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00014434208938412648,
+      "loss": 2.3242,
+      "step": 8235
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00014430186250452554,
+      "loss": 1.712,
+      "step": 8240
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001442616356249246,
+      "loss": 1.66,
+      "step": 8245
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00014422140874532363,
+      "loss": 1.6371,
+      "step": 8250
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00014418118186572268,
+      "loss": 4.1268,
+      "step": 8255
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00014414095498612172,
+      "loss": 4.1861,
+      "step": 8260
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00014410072810652077,
+      "loss": 3.8643,
+      "step": 8265
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00014406050122691983,
+      "loss": 3.8318,
+      "step": 8270
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001440202743473189,
+      "loss": 3.3503,
+      "step": 8275
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00014398004746771795,
+      "loss": 3.3456,
+      "step": 8280
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00014393982058811698,
+      "loss": 4.1158,
+      "step": 8285
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00014389959370851604,
+      "loss": 3.0195,
+      "step": 8290
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001438593668289151,
+      "loss": 0.8465,
+      "step": 8295
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00014381913994931413,
+      "loss": 1.5685,
+      "step": 8300
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00014377891306971319,
+      "loss": 4.1578,
+      "step": 8305
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014373868619011224,
+      "loss": 4.6133,
+      "step": 8310
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001436984593105113,
+      "loss": 5.6063,
+      "step": 8315
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014365823243091033,
+      "loss": 4.8566,
+      "step": 8320
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001436180055513094,
+      "loss": 3.8412,
+      "step": 8325
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014357777867170845,
+      "loss": 2.9457,
+      "step": 8330
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014353755179210748,
+      "loss": 3.2986,
+      "step": 8335
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014349732491250654,
+      "loss": 3.7619,
+      "step": 8340
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001434570980329056,
+      "loss": 2.0469,
+      "step": 8345
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014341687115330466,
+      "loss": 1.2048,
+      "step": 8350
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001433766442737037,
+      "loss": 4.6137,
+      "step": 8355
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014333641739410274,
+      "loss": 4.2262,
+      "step": 8360
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001432961905145018,
+      "loss": 3.5321,
+      "step": 8365
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014325596363490086,
+      "loss": 3.1123,
+      "step": 8370
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001432157367552999,
+      "loss": 4.0621,
+      "step": 8375
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014317550987569895,
+      "loss": 3.2757,
+      "step": 8380
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000143135282996098,
+      "loss": 3.4299,
+      "step": 8385
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014309505611649704,
+      "loss": 2.9666,
+      "step": 8390
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001430548292368961,
+      "loss": 3.7125,
+      "step": 8395
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014301460235729516,
+      "loss": 3.3482,
+      "step": 8400
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014297437547769421,
+      "loss": 3.2589,
+      "step": 8405
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014293414859809325,
+      "loss": 4.1152,
+      "step": 8410
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001428939217184923,
+      "loss": 4.1553,
+      "step": 8415
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014285369483889136,
+      "loss": 2.563,
+      "step": 8420
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001428134679592904,
+      "loss": 3.1831,
+      "step": 8425
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014277324107968945,
+      "loss": 3.0423,
+      "step": 8430
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001427330142000885,
+      "loss": 2.415,
+      "step": 8435
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00014269278732048757,
+      "loss": 3.3074,
+      "step": 8440
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00014265256044088663,
+      "loss": 2.9256,
+      "step": 8445
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00014261233356128566,
+      "loss": 0.7464,
+      "step": 8450
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00014257210668168472,
+      "loss": 3.7371,
+      "step": 8455
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00014253187980208375,
+      "loss": 3.3926,
+      "step": 8460
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001424916529224828,
+      "loss": 3.8283,
+      "step": 8465
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00014245142604288186,
+      "loss": 2.7588,
+      "step": 8470
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00014241119916328092,
+      "loss": 3.9244,
+      "step": 8475
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00014237097228367998,
+      "loss": 3.9608,
+      "step": 8480
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000142330745404079,
+      "loss": 3.4815,
+      "step": 8485
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00014229051852447804,
+      "loss": 2.2937,
+      "step": 8490
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001422502916448771,
+      "loss": 1.4673,
+      "step": 8495
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00014221811014119635,
+      "loss": 2.0945,
+      "step": 8500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001421778832615954,
+      "loss": 5.1734,
+      "step": 8505
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00014213765638199446,
+      "loss": 4.7025,
+      "step": 8510
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00014209742950239352,
+      "loss": 3.0898,
+      "step": 8515
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00014205720262279255,
+      "loss": 3.6896,
+      "step": 8520
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001420169757431916,
+      "loss": 3.1475,
+      "step": 8525
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00014197674886359064,
+      "loss": 1.9613,
+      "step": 8530
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001419365219839897,
+      "loss": 2.6826,
+      "step": 8535
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00014189629510438876,
+      "loss": 1.0485,
+      "step": 8540
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00014185606822478782,
+      "loss": 2.8809,
+      "step": 8545
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00014181584134518685,
+      "loss": 0.0693,
+      "step": 8550
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001417756144655859,
+      "loss": 4.4512,
+      "step": 8555
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00014173538758598496,
+      "loss": 4.483,
+      "step": 8560
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00014169516070638402,
+      "loss": 4.1109,
+      "step": 8565
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00014165493382678305,
+      "loss": 3.7912,
+      "step": 8570
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001416147069471821,
+      "loss": 4.1405,
+      "step": 8575
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00014157448006758117,
+      "loss": 2.772,
+      "step": 8580
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001415342531879802,
+      "loss": 2.3251,
+      "step": 8585
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00014149402630837926,
+      "loss": 3.733,
+      "step": 8590
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00014145379942877832,
+      "loss": 1.8271,
+      "step": 8595
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00014141357254917738,
+      "loss": 1.507,
+      "step": 8600
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001413733456695764,
+      "loss": 5.0523,
+      "step": 8605
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00014133311878997547,
+      "loss": 4.0752,
+      "step": 8610
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00014129289191037452,
+      "loss": 4.1002,
+      "step": 8615
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00014125266503077356,
+      "loss": 3.0717,
+      "step": 8620
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001412124381511726,
+      "loss": 2.2346,
+      "step": 8625
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00014117221127157167,
+      "loss": 2.3687,
+      "step": 8630
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00014113198439197073,
+      "loss": 2.5631,
+      "step": 8635
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001410917575123698,
+      "loss": 1.2662,
+      "step": 8640
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00014105153063276882,
+      "loss": 3.2849,
+      "step": 8645
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00014101130375316788,
+      "loss": 1.8528,
+      "step": 8650
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001409710768735669,
+      "loss": 4.3492,
+      "step": 8655
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00014093084999396597,
+      "loss": 4.6064,
+      "step": 8660
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00014089062311436503,
+      "loss": 4.0658,
+      "step": 8665
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00014085039623476408,
+      "loss": 3.9941,
+      "step": 8670
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00014081016935516314,
+      "loss": 3.4873,
+      "step": 8675
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00014076994247556217,
+      "loss": 1.615,
+      "step": 8680
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00014072971559596123,
+      "loss": 1.9473,
+      "step": 8685
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00014068948871636026,
+      "loss": 3.5422,
+      "step": 8690
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00014064926183675932,
+      "loss": 1.6255,
+      "step": 8695
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00014060903495715838,
+      "loss": 0.6512,
+      "step": 8700
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00014056880807755744,
+      "loss": 4.1771,
+      "step": 8705
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001405285811979565,
+      "loss": 4.2812,
+      "step": 8710
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00014048835431835555,
+      "loss": 3.3185,
+      "step": 8715
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00014044812743875458,
+      "loss": 3.8672,
+      "step": 8720
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00014040790055915362,
+      "loss": 4.4287,
+      "step": 8725
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00014036767367955267,
+      "loss": 3.4561,
+      "step": 8730
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00014032744679995173,
+      "loss": 3.5,
+      "step": 8735
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001402872199203508,
+      "loss": 2.2681,
+      "step": 8740
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00014024699304074985,
+      "loss": 1.6788,
+      "step": 8745
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001402067661611489,
+      "loss": 4.0663,
+      "step": 8750
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00014016653928154794,
+      "loss": 4.5193,
+      "step": 8755
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00014012631240194697,
+      "loss": 4.5533,
+      "step": 8760
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00014008608552234603,
+      "loss": 3.6523,
+      "step": 8765
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00014004585864274509,
+      "loss": 4.5703,
+      "step": 8770
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00014000563176314414,
+      "loss": 4.8623,
+      "step": 8775
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001399654048835432,
+      "loss": 3.0412,
+      "step": 8780
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00013992517800394226,
+      "loss": 2.1579,
+      "step": 8785
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001398849511243413,
+      "loss": 3.1091,
+      "step": 8790
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00013984472424474032,
+      "loss": 2.0995,
+      "step": 8795
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00013980449736513938,
+      "loss": 2.2463,
+      "step": 8800
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00013976427048553844,
+      "loss": 5.2713,
+      "step": 8805
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001397240436059375,
+      "loss": 3.9492,
+      "step": 8810
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00013968381672633656,
+      "loss": 3.9627,
+      "step": 8815
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00013964358984673561,
+      "loss": 4.1762,
+      "step": 8820
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00013960336296713465,
+      "loss": 3.9076,
+      "step": 8825
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00013956313608753368,
+      "loss": 2.9175,
+      "step": 8830
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00013952290920793273,
+      "loss": 1.738,
+      "step": 8835
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001394826823283318,
+      "loss": 1.9772,
+      "step": 8840
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00013944245544873085,
+      "loss": 1.0877,
+      "step": 8845
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001394022285691299,
+      "loss": 1.9826,
+      "step": 8850
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00013936200168952894,
+      "loss": 5.7988,
+      "step": 8855
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000139321774809928,
+      "loss": 5.0049,
+      "step": 8860
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00013928154793032706,
+      "loss": 4.1568,
+      "step": 8865
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001392413210507261,
+      "loss": 3.3445,
+      "step": 8870
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00013920109417112515,
+      "loss": 4.5633,
+      "step": 8875
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001391608672915242,
+      "loss": 2.7033,
+      "step": 8880
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00013912064041192326,
+      "loss": 3.8346,
+      "step": 8885
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001390804135323223,
+      "loss": 2.3505,
+      "step": 8890
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00013904018665272135,
+      "loss": 1.7089,
+      "step": 8895
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001389999597731204,
+      "loss": 1.9751,
+      "step": 8900
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00013895973289351944,
+      "loss": 3.9482,
+      "step": 8905
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001389195060139185,
+      "loss": 5.2367,
+      "step": 8910
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00013887927913431756,
+      "loss": 4.2158,
+      "step": 8915
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00013883905225471662,
+      "loss": 4.0254,
+      "step": 8920
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00013879882537511565,
+      "loss": 4.4863,
+      "step": 8925
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001387585984955147,
+      "loss": 3.8158,
+      "step": 8930
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00013871837161591376,
+      "loss": 2.6618,
+      "step": 8935
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00013867814473631282,
+      "loss": 2.6897,
+      "step": 8940
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00013863791785671185,
+      "loss": 1.2268,
+      "step": 8945
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001385976909771109,
+      "loss": 2.3218,
+      "step": 8950
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00013855746409750997,
+      "loss": 3.4869,
+      "step": 8955
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000138517237217909,
+      "loss": 4.4197,
+      "step": 8960
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013847701033830806,
+      "loss": 3.4383,
+      "step": 8965
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013843678345870712,
+      "loss": 3.1334,
+      "step": 8970
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013839655657910618,
+      "loss": 3.8105,
+      "step": 8975
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001383563296995052,
+      "loss": 2.6163,
+      "step": 8980
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013831610281990426,
+      "loss": 2.1495,
+      "step": 8985
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013827587594030332,
+      "loss": 3.512,
+      "step": 8990
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013823564906070235,
+      "loss": 4.8705,
+      "step": 8995
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001381954221811014,
+      "loss": 1.2528,
+      "step": 9000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013815519530150047,
+      "loss": 4.6674,
+      "step": 9005
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013811496842189953,
+      "loss": 3.8508,
+      "step": 9010
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001380747415422986,
+      "loss": 3.9963,
+      "step": 9015
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013803451466269762,
+      "loss": 4.5911,
+      "step": 9020
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013799428778309668,
+      "loss": 4.2135,
+      "step": 9025
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001379540609034957,
+      "loss": 3.3,
+      "step": 9030
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013791383402389477,
+      "loss": 3.1516,
+      "step": 9035
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013787360714429382,
+      "loss": 2.2726,
+      "step": 9040
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013783338026469288,
+      "loss": 1.3781,
+      "step": 9045
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013779315338509194,
+      "loss": 1.5878,
+      "step": 9050
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013775292650549097,
+      "loss": 3.8682,
+      "step": 9055
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013771269962589003,
+      "loss": 5.8621,
+      "step": 9060
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013767247274628906,
+      "loss": 4.1668,
+      "step": 9065
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013763224586668812,
+      "loss": 3.3687,
+      "step": 9070
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013759201898708718,
+      "loss": 3.5014,
+      "step": 9075
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013755179210748624,
+      "loss": 2.693,
+      "step": 9080
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001375115652278853,
+      "loss": 3.4979,
+      "step": 9085
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013747133834828435,
+      "loss": 2.3423,
+      "step": 9090
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00013743111146868338,
+      "loss": 3.0961,
+      "step": 9095
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00013739088458908241,
+      "loss": 2.2662,
+      "step": 9100
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00013735065770948147,
+      "loss": 4.8318,
+      "step": 9105
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00013731043082988053,
+      "loss": 4.9508,
+      "step": 9110
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001372702039502796,
+      "loss": 3.5961,
+      "step": 9115
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00013722997707067865,
+      "loss": 4.1072,
+      "step": 9120
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001371897501910777,
+      "loss": 3.7145,
+      "step": 9125
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00013714952331147674,
+      "loss": 3.7985,
+      "step": 9130
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00013710929643187577,
+      "loss": 3.2468,
+      "step": 9135
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00013706906955227483,
+      "loss": 2.0802,
+      "step": 9140
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00013702884267267388,
+      "loss": 1.4174,
+      "step": 9145
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00013698861579307294,
+      "loss": 0.3039,
+      "step": 9150
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.000136948388913472,
+      "loss": 4.1092,
+      "step": 9155
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00013690816203387103,
+      "loss": 4.8393,
+      "step": 9160
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001368679351542701,
+      "loss": 4.3215,
+      "step": 9165
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00013682770827466912,
+      "loss": 4.8441,
+      "step": 9170
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00013678748139506818,
+      "loss": 3.8361,
+      "step": 9175
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00013674725451546724,
+      "loss": 3.663,
+      "step": 9180
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001367070276358663,
+      "loss": 3.2722,
+      "step": 9185
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00013666680075626535,
+      "loss": 3.3886,
+      "step": 9190
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00013662657387666439,
+      "loss": 3.5016,
+      "step": 9195
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00013658634699706344,
+      "loss": 2.5376,
+      "step": 9200
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001365461201174625,
+      "loss": 4.8418,
+      "step": 9205
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00013650589323786153,
+      "loss": 4.2836,
+      "step": 9210
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001364656663582606,
+      "loss": 4.0514,
+      "step": 9215
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00013642543947865965,
+      "loss": 3.6088,
+      "step": 9220
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001363852125990587,
+      "loss": 3.5259,
+      "step": 9225
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00013634498571945774,
+      "loss": 3.9555,
+      "step": 9230
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001363047588398568,
+      "loss": 2.5013,
+      "step": 9235
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00013626453196025586,
+      "loss": 1.8686,
+      "step": 9240
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001362243050806549,
+      "loss": 2.7858,
+      "step": 9245
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00013618407820105395,
+      "loss": 2.5493,
+      "step": 9250
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.000136143851321453,
+      "loss": 4.5645,
+      "step": 9255
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00013610362444185206,
+      "loss": 4.4248,
+      "step": 9260
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001360633975622511,
+      "loss": 4.1404,
+      "step": 9265
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00013602317068265015,
+      "loss": 4.19,
+      "step": 9270
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001359829438030492,
+      "loss": 3.1199,
+      "step": 9275
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00013594271692344827,
+      "loss": 3.3325,
+      "step": 9280
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001359024900438473,
+      "loss": 1.8506,
+      "step": 9285
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00013586226316424636,
+      "loss": 2.0855,
+      "step": 9290
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00013582203628464542,
+      "loss": 2.7468,
+      "step": 9295
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00013578180940504445,
+      "loss": 3.2028,
+      "step": 9300
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001357415825254435,
+      "loss": 4.6888,
+      "step": 9305
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00013570135564584256,
+      "loss": 4.9221,
+      "step": 9310
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00013566112876624162,
+      "loss": 4.0654,
+      "step": 9315
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00013562090188664065,
+      "loss": 3.5768,
+      "step": 9320
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001355806750070397,
+      "loss": 4.2205,
+      "step": 9325
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00013554044812743877,
+      "loss": 2.9903,
+      "step": 9330
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001355002212478378,
+      "loss": 3.491,
+      "step": 9335
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00013545999436823686,
+      "loss": 2.4792,
+      "step": 9340
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00013541976748863592,
+      "loss": 0.8765,
+      "step": 9345
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00013537954060903497,
+      "loss": 1.7416,
+      "step": 9350
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00013533931372943403,
+      "loss": 4.7154,
+      "step": 9355
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00013529908684983306,
+      "loss": 3.5176,
+      "step": 9360
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00013525885997023212,
+      "loss": 3.8848,
+      "step": 9365
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00013521863309063115,
+      "loss": 2.9452,
+      "step": 9370
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001351784062110302,
+      "loss": 3.007,
+      "step": 9375
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00013513817933142927,
+      "loss": 3.152,
+      "step": 9380
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00013509795245182833,
+      "loss": 3.2887,
+      "step": 9385
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00013505772557222739,
+      "loss": 2.8927,
+      "step": 9390
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00013501749869262642,
+      "loss": 2.6005,
+      "step": 9395
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00013497727181302548,
+      "loss": 3.1346,
+      "step": 9400
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001349370449334245,
+      "loss": 4.4902,
+      "step": 9405
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00013489681805382357,
+      "loss": 4.3199,
+      "step": 9410
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00013485659117422262,
+      "loss": 4.7805,
+      "step": 9415
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00013481636429462168,
+      "loss": 3.8885,
+      "step": 9420
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00013477613741502074,
+      "loss": 2.4693,
+      "step": 9425
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001347359105354198,
+      "loss": 2.7459,
+      "step": 9430
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00013469568365581883,
+      "loss": 3.4011,
+      "step": 9435
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00013465545677621786,
+      "loss": 3.0815,
+      "step": 9440
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00013461522989661692,
+      "loss": 1.5173,
+      "step": 9445
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00013457500301701598,
+      "loss": 2.4051,
+      "step": 9450
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00013453477613741504,
+      "loss": 4.7314,
+      "step": 9455
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001344945492578141,
+      "loss": 4.0549,
+      "step": 9460
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00013445432237821312,
+      "loss": 3.774,
+      "step": 9465
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00013441409549861218,
+      "loss": 3.5994,
+      "step": 9470
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00013437386861901121,
+      "loss": 3.5275,
+      "step": 9475
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00013433364173941027,
+      "loss": 3.8396,
+      "step": 9480
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00013429341485980933,
+      "loss": 2.5812,
+      "step": 9485
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001342531879802084,
+      "loss": 3.3454,
+      "step": 9490
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00013421296110060745,
+      "loss": 1.5726,
+      "step": 9495
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00013417273422100648,
+      "loss": 0.6411,
+      "step": 9500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00013413250734140554,
+      "loss": 4.6568,
+      "step": 9505
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00013409228046180457,
+      "loss": 4.666,
+      "step": 9510
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00013405205358220363,
+      "loss": 3.4013,
+      "step": 9515
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00013401182670260268,
+      "loss": 3.9559,
+      "step": 9520
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00013397159982300174,
+      "loss": 3.7091,
+      "step": 9525
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001339313729434008,
+      "loss": 3.2853,
+      "step": 9530
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00013389114606379983,
+      "loss": 2.6608,
+      "step": 9535
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001338509191841989,
+      "loss": 1.7764,
+      "step": 9540
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00013381069230459792,
+      "loss": 2.1373,
+      "step": 9545
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00013377046542499698,
+      "loss": 2.9499,
+      "step": 9550
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00013373023854539604,
+      "loss": 4.3037,
+      "step": 9555
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001336900116657951,
+      "loss": 3.468,
+      "step": 9560
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00013364978478619415,
+      "loss": 4.5646,
+      "step": 9565
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00013360955790659318,
+      "loss": 3.7449,
+      "step": 9570
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00013356933102699224,
+      "loss": 3.6267,
+      "step": 9575
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001335291041473913,
+      "loss": 3.0011,
+      "step": 9580
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00013348887726779033,
+      "loss": 3.9645,
+      "step": 9585
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001334486503881894,
+      "loss": 2.6243,
+      "step": 9590
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00013340842350858845,
+      "loss": 3.0731,
+      "step": 9595
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001333681966289875,
+      "loss": 1.45,
+      "step": 9600
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00013332796974938654,
+      "loss": 3.6967,
+      "step": 9605
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001332877428697856,
+      "loss": 4.2301,
+      "step": 9610
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00013324751599018465,
+      "loss": 3.9514,
+      "step": 9615
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00013320728911058369,
+      "loss": 3.9246,
+      "step": 9620
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00013316706223098274,
+      "loss": 3.4285,
+      "step": 9625
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001331268353513818,
+      "loss": 3.0667,
+      "step": 9630
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00013308660847178086,
+      "loss": 2.9836,
+      "step": 9635
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001330463815921799,
+      "loss": 2.4699,
+      "step": 9640
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00013300615471257895,
+      "loss": 2.3858,
+      "step": 9645
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.000132965927832978,
+      "loss": 1.9694,
+      "step": 9650
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00013292570095337707,
+      "loss": 4.6979,
+      "step": 9655
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001328854740737761,
+      "loss": 4.1387,
+      "step": 9660
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00013284524719417516,
+      "loss": 4.4168,
+      "step": 9665
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00013280502031457421,
+      "loss": 4.8387,
+      "step": 9670
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00013276479343497325,
+      "loss": 3.6943,
+      "step": 9675
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001327245665553723,
+      "loss": 3.0823,
+      "step": 9680
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00013268433967577136,
+      "loss": 2.7571,
+      "step": 9685
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00013264411279617042,
+      "loss": 3.0168,
+      "step": 9690
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00013260388591656945,
+      "loss": 2.7996,
+      "step": 9695
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001325636590369685,
+      "loss": 2.6553,
+      "step": 9700
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00013252343215736757,
+      "loss": 4.643,
+      "step": 9705
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001324832052777666,
+      "loss": 4.3184,
+      "step": 9710
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00013244297839816566,
+      "loss": 3.1918,
+      "step": 9715
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00013240275151856472,
+      "loss": 2.6649,
+      "step": 9720
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00013236252463896377,
+      "loss": 3.9285,
+      "step": 9725
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00013232229775936283,
+      "loss": 2.9876,
+      "step": 9730
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00013228207087976186,
+      "loss": 2.318,
+      "step": 9735
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00013224184400016092,
+      "loss": 2.6796,
+      "step": 9740
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00013220161712055995,
+      "loss": 1.675,
+      "step": 9745
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.000132161390240959,
+      "loss": 1.1842,
+      "step": 9750
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00013212116336135807,
+      "loss": 3.9574,
+      "step": 9755
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00013208093648175713,
+      "loss": 5.1955,
+      "step": 9760
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00013204070960215619,
+      "loss": 3.4871,
+      "step": 9765
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00013200048272255522,
+      "loss": 4.4609,
+      "step": 9770
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00013196025584295427,
+      "loss": 5.2233,
+      "step": 9775
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001319200289633533,
+      "loss": 4.0359,
+      "step": 9780
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00013187980208375236,
+      "loss": 2.6144,
+      "step": 9785
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00013183957520415142,
+      "loss": 1.9424,
+      "step": 9790
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00013179934832455048,
+      "loss": 3.2209,
+      "step": 9795
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00013175912144494954,
+      "loss": 2.7711,
+      "step": 9800
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00013171889456534857,
+      "loss": 3.8485,
+      "step": 9805
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001316786676857476,
+      "loss": 3.1569,
+      "step": 9810
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00013163844080614666,
+      "loss": 4.6219,
+      "step": 9815
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00013159821392654572,
+      "loss": 3.0187,
+      "step": 9820
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00013155798704694478,
+      "loss": 3.7212,
+      "step": 9825
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00013151776016734383,
+      "loss": 2.5,
+      "step": 9830
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001314775332877429,
+      "loss": 2.1953,
+      "step": 9835
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00013143730640814192,
+      "loss": 1.6314,
+      "step": 9840
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00013139707952854095,
+      "loss": 2.1152,
+      "step": 9845
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00013135685264894,
+      "loss": 3.3051,
+      "step": 9850
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00013131662576933907,
+      "loss": 4.8615,
+      "step": 9855
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00013127639888973813,
+      "loss": 4.4182,
+      "step": 9860
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001312361720101372,
+      "loss": 4.64,
+      "step": 9865
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00013119594513053625,
+      "loss": 3.8684,
+      "step": 9870
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00013115571825093528,
+      "loss": 3.9086,
+      "step": 9875
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013111549137133434,
+      "loss": 2.4209,
+      "step": 9880
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013107526449173337,
+      "loss": 1.358,
+      "step": 9885
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013103503761213242,
+      "loss": 1.3452,
+      "step": 9890
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013099481073253148,
+      "loss": 2.1419,
+      "step": 9895
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013095458385293054,
+      "loss": 2.2326,
+      "step": 9900
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001309143569733296,
+      "loss": 4.982,
+      "step": 9905
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013087413009372863,
+      "loss": 4.2432,
+      "step": 9910
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001308339032141277,
+      "loss": 4.2863,
+      "step": 9915
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013079367633452672,
+      "loss": 3.5053,
+      "step": 9920
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013075344945492578,
+      "loss": 4.0033,
+      "step": 9925
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013071322257532484,
+      "loss": 3.812,
+      "step": 9930
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001306729956957239,
+      "loss": 3.1587,
+      "step": 9935
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013063276881612295,
+      "loss": 3.0073,
+      "step": 9940
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013059254193652198,
+      "loss": 3.4743,
+      "step": 9945
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013055231505692104,
+      "loss": 1.0452,
+      "step": 9950
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001305120881773201,
+      "loss": 3.9975,
+      "step": 9955
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013047186129771913,
+      "loss": 3.7115,
+      "step": 9960
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001304316344181182,
+      "loss": 3.8148,
+      "step": 9965
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013039140753851725,
+      "loss": 3.2805,
+      "step": 9970
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001303511806589163,
+      "loss": 3.5577,
+      "step": 9975
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013031095377931534,
+      "loss": 3.2957,
+      "step": 9980
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001302707268997144,
+      "loss": 1.8262,
+      "step": 9985
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013023050002011345,
+      "loss": 3.8027,
+      "step": 9990
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013019027314051249,
+      "loss": 1.8096,
+      "step": 9995
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013015004626091154,
+      "loss": 1.4512,
+      "step": 10000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001301098193813106,
+      "loss": 3.9459,
+      "step": 10005
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013006959250170966,
+      "loss": 4.8336,
+      "step": 10010
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001300293656221087,
+      "loss": 4.3008,
+      "step": 10015
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012998913874250775,
+      "loss": 3.5926,
+      "step": 10020
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001299489118629068,
+      "loss": 3.3028,
+      "step": 10025
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012990868498330587,
+      "loss": 3.0276,
+      "step": 10030
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001298684581037049,
+      "loss": 1.6533,
+      "step": 10035
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012982823122410396,
+      "loss": 2.652,
+      "step": 10040
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.000129788004344503,
+      "loss": 1.1642,
+      "step": 10045
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012974777746490204,
+      "loss": 2.6869,
+      "step": 10050
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001297075505853011,
+      "loss": 4.8092,
+      "step": 10055
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012966732370570016,
+      "loss": 5.8297,
+      "step": 10060
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012962709682609922,
+      "loss": 3.7469,
+      "step": 10065
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012958686994649825,
+      "loss": 4.5357,
+      "step": 10070
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001295466430668973,
+      "loss": 3.599,
+      "step": 10075
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012950641618729634,
+      "loss": 2.9362,
+      "step": 10080
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001294661893076954,
+      "loss": 2.9291,
+      "step": 10085
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012942596242809446,
+      "loss": 0.7292,
+      "step": 10090
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012938573554849351,
+      "loss": 2.6529,
+      "step": 10095
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012934550866889257,
+      "loss": 1.4879,
+      "step": 10100
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012930528178929163,
+      "loss": 4.5059,
+      "step": 10105
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012926505490969066,
+      "loss": 4.3889,
+      "step": 10110
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001292248280300897,
+      "loss": 5.0908,
+      "step": 10115
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012918460115048875,
+      "loss": 4.3072,
+      "step": 10120
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001291443742708878,
+      "loss": 3.3316,
+      "step": 10125
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012910414739128687,
+      "loss": 3.8155,
+      "step": 10130
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012906392051168593,
+      "loss": 2.2591,
+      "step": 10135
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012902369363208498,
+      "loss": 3.473,
+      "step": 10140
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012898346675248402,
+      "loss": 1.5188,
+      "step": 10145
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012894323987288305,
+      "loss": 2.933,
+      "step": 10150
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001289030129932821,
+      "loss": 4.2186,
+      "step": 10155
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012886278611368116,
+      "loss": 4.9096,
+      "step": 10160
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012882255923408022,
+      "loss": 4.5613,
+      "step": 10165
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012878233235447928,
+      "loss": 3.0895,
+      "step": 10170
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012874210547487834,
+      "loss": 3.5822,
+      "step": 10175
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012870187859527737,
+      "loss": 2.599,
+      "step": 10180
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001286616517156764,
+      "loss": 3.6651,
+      "step": 10185
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012862142483607546,
+      "loss": 2.747,
+      "step": 10190
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012858119795647452,
+      "loss": 2.2399,
+      "step": 10195
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012854097107687357,
+      "loss": 0.2028,
+      "step": 10200
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012850074419727263,
+      "loss": 3.8363,
+      "step": 10205
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001284605173176717,
+      "loss": 3.6502,
+      "step": 10210
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012842029043807072,
+      "loss": 4.1717,
+      "step": 10215
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012838006355846975,
+      "loss": 3.6252,
+      "step": 10220
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001283398366788688,
+      "loss": 3.8029,
+      "step": 10225
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012829960979926787,
+      "loss": 2.7107,
+      "step": 10230
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012825938291966693,
+      "loss": 3.614,
+      "step": 10235
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.000128219156040066,
+      "loss": 2.4932,
+      "step": 10240
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012817892916046504,
+      "loss": 3.0111,
+      "step": 10245
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012813870228086408,
+      "loss": 2.2005,
+      "step": 10250
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012809847540126313,
+      "loss": 4.8871,
+      "step": 10255
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012805824852166217,
+      "loss": 4.4473,
+      "step": 10260
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012801802164206122,
+      "loss": 4.5402,
+      "step": 10265
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012797779476246028,
+      "loss": 3.0773,
+      "step": 10270
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00012793756788285934,
+      "loss": 2.9697,
+      "step": 10275
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001278973410032584,
+      "loss": 1.9987,
+      "step": 10280
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00012785711412365743,
+      "loss": 3.0007,
+      "step": 10285
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001278168872440565,
+      "loss": 2.0372,
+      "step": 10290
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00012777666036445552,
+      "loss": 2.4596,
+      "step": 10295
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00012773643348485458,
+      "loss": 1.395,
+      "step": 10300
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00012769620660525364,
+      "loss": 4.6918,
+      "step": 10305
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001276559797256527,
+      "loss": 3.618,
+      "step": 10310
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00012761575284605175,
+      "loss": 3.8582,
+      "step": 10315
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00012757552596645078,
+      "loss": 4.6336,
+      "step": 10320
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00012753529908684984,
+      "loss": 2.6338,
+      "step": 10325
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001274950722072489,
+      "loss": 3.15,
+      "step": 10330
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00012745484532764793,
+      "loss": 1.2446,
+      "step": 10335
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.000127414618448047,
+      "loss": 2.2504,
+      "step": 10340
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00012737439156844605,
+      "loss": 3.426,
+      "step": 10345
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001273341646888451,
+      "loss": 0.0367,
+      "step": 10350
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00012729393780924414,
+      "loss": 4.5221,
+      "step": 10355
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001272537109296432,
+      "loss": 4.7127,
+      "step": 10360
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00012721348405004225,
+      "loss": 3.3664,
+      "step": 10365
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00012717325717044128,
+      "loss": 3.905,
+      "step": 10370
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00012713303029084034,
+      "loss": 3.6577,
+      "step": 10375
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001270928034112394,
+      "loss": 2.1809,
+      "step": 10380
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00012705257653163843,
+      "loss": 2.0813,
+      "step": 10385
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001270123496520375,
+      "loss": 3.2018,
+      "step": 10390
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00012697212277243655,
+      "loss": 2.3257,
+      "step": 10395
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001269318958928356,
+      "loss": 1.8804,
+      "step": 10400
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00012689166901323466,
+      "loss": 4.5512,
+      "step": 10405
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001268514421336337,
+      "loss": 3.9561,
+      "step": 10410
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00012681121525403275,
+      "loss": 3.5211,
+      "step": 10415
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00012677098837443179,
+      "loss": 3.8664,
+      "step": 10420
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00012673076149483084,
+      "loss": 3.3927,
+      "step": 10425
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001266905346152299,
+      "loss": 3.9993,
+      "step": 10430
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00012665030773562896,
+      "loss": 3.6185,
+      "step": 10435
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00012661008085602802,
+      "loss": 2.7498,
+      "step": 10440
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00012656985397642705,
+      "loss": 1.6683,
+      "step": 10445
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001265296270968261,
+      "loss": 0.0475,
+      "step": 10450
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00012648940021722514,
+      "loss": 4.1752,
+      "step": 10455
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001264491733376242,
+      "loss": 4.8688,
+      "step": 10460
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00012640894645802326,
+      "loss": 4.3963,
+      "step": 10465
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001263687195784223,
+      "loss": 4.7291,
+      "step": 10470
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00012632849269882137,
+      "loss": 3.6086,
+      "step": 10475
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00012628826581922043,
+      "loss": 2.9757,
+      "step": 10480
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00012624803893961946,
+      "loss": 3.5504,
+      "step": 10485
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001262078120600185,
+      "loss": 3.5773,
+      "step": 10490
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00012616758518041755,
+      "loss": 1.6866,
+      "step": 10495
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001261273583008166,
+      "loss": 2.5525,
+      "step": 10500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00012608713142121567,
+      "loss": 4.3145,
+      "step": 10505
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00012604690454161473,
+      "loss": 4.4189,
+      "step": 10510
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00012600667766201378,
+      "loss": 4.108,
+      "step": 10515
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00012596645078241281,
+      "loss": 4.3092,
+      "step": 10520
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00012592622390281185,
+      "loss": 4.0993,
+      "step": 10525
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001258859970232109,
+      "loss": 3.2504,
+      "step": 10530
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00012584577014360996,
+      "loss": 3.6479,
+      "step": 10535
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00012580554326400902,
+      "loss": 4.1616,
+      "step": 10540
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00012577336176032827,
+      "loss": 3.0753,
+      "step": 10545
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001257331348807273,
+      "loss": 1.1912,
+      "step": 10550
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00012569290800112636,
+      "loss": 4.1457,
+      "step": 10555
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00012565268112152541,
+      "loss": 4.3012,
+      "step": 10560
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00012561245424192447,
+      "loss": 3.3361,
+      "step": 10565
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001255722273623235,
+      "loss": 3.3066,
+      "step": 10570
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00012553200048272256,
+      "loss": 3.2586,
+      "step": 10575
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00012549177360312162,
+      "loss": 3.1351,
+      "step": 10580
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00012545154672352065,
+      "loss": 3.3842,
+      "step": 10585
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001254113198439197,
+      "loss": 2.4289,
+      "step": 10590
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00012537109296431877,
+      "loss": 2.9674,
+      "step": 10595
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00012533086608471783,
+      "loss": 1.956,
+      "step": 10600
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00012529063920511686,
+      "loss": 3.8779,
+      "step": 10605
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00012525041232551592,
+      "loss": 3.518,
+      "step": 10610
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00012521018544591497,
+      "loss": 4.851,
+      "step": 10615
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.000125169958566314,
+      "loss": 3.1552,
+      "step": 10620
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00012512973168671306,
+      "loss": 2.9098,
+      "step": 10625
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00012508950480711212,
+      "loss": 2.8324,
+      "step": 10630
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00012504927792751118,
+      "loss": 2.5265,
+      "step": 10635
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00012500905104791024,
+      "loss": 2.6673,
+      "step": 10640
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00012496882416830927,
+      "loss": 2.8797,
+      "step": 10645
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00012492859728870833,
+      "loss": 2.8138,
+      "step": 10650
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00012488837040910736,
+      "loss": 4.6326,
+      "step": 10655
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00012484814352950642,
+      "loss": 3.6939,
+      "step": 10660
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00012480791664990548,
+      "loss": 3.8156,
+      "step": 10665
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00012476768977030453,
+      "loss": 3.8127,
+      "step": 10670
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001247274628907036,
+      "loss": 3.8121,
+      "step": 10675
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00012468723601110262,
+      "loss": 4.3749,
+      "step": 10680
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00012464700913150168,
+      "loss": 2.2904,
+      "step": 10685
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001246067822519007,
+      "loss": 1.6336,
+      "step": 10690
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00012456655537229977,
+      "loss": 1.492,
+      "step": 10695
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00012452632849269883,
+      "loss": 1.4875,
+      "step": 10700
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001244861016130979,
+      "loss": 4.3963,
+      "step": 10705
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00012444587473349695,
+      "loss": 3.8414,
+      "step": 10710
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.000124405647853896,
+      "loss": 4.0334,
+      "step": 10715
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00012436542097429503,
+      "loss": 4.182,
+      "step": 10720
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00012432519409469407,
+      "loss": 3.2245,
+      "step": 10725
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00012428496721509312,
+      "loss": 2.7608,
+      "step": 10730
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00012424474033549218,
+      "loss": 2.8074,
+      "step": 10735
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00012420451345589124,
+      "loss": 3.0531,
+      "step": 10740
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001241642865762903,
+      "loss": 1.4641,
+      "step": 10745
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00012412405969668933,
+      "loss": 2.0748,
+      "step": 10750
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001240838328170884,
+      "loss": 4.573,
+      "step": 10755
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00012404360593748742,
+      "loss": 3.651,
+      "step": 10760
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00012400337905788648,
+      "loss": 3.5417,
+      "step": 10765
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00012396315217828554,
+      "loss": 4.6764,
+      "step": 10770
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001239229252986846,
+      "loss": 4.1459,
+      "step": 10775
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00012388269841908365,
+      "loss": 3.444,
+      "step": 10780
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00012384247153948268,
+      "loss": 3.3812,
+      "step": 10785
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00012380224465988174,
+      "loss": 1.5143,
+      "step": 10790
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00012376201778028077,
+      "loss": 2.626,
+      "step": 10795
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00012372179090067983,
+      "loss": 1.1325,
+      "step": 10800
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001236815640210789,
+      "loss": 5.0693,
+      "step": 10805
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00012364133714147795,
+      "loss": 4.8133,
+      "step": 10810
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.000123601110261877,
+      "loss": 3.3014,
+      "step": 10815
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00012356088338227604,
+      "loss": 4.2463,
+      "step": 10820
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001235206565026751,
+      "loss": 4.0414,
+      "step": 10825
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00012348042962307413,
+      "loss": 4.8287,
+      "step": 10830
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00012344020274347318,
+      "loss": 1.6694,
+      "step": 10835
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00012339997586387224,
+      "loss": 1.8969,
+      "step": 10840
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001233597489842713,
+      "loss": 2.2986,
+      "step": 10845
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00012331952210467036,
+      "loss": 1.7887,
+      "step": 10850
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001232792952250694,
+      "loss": 4.7313,
+      "step": 10855
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00012323906834546845,
+      "loss": 3.7457,
+      "step": 10860
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001231988414658675,
+      "loss": 5.1592,
+      "step": 10865
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00012315861458626654,
+      "loss": 3.3604,
+      "step": 10870
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001231183877066656,
+      "loss": 3.1216,
+      "step": 10875
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00012307816082706465,
+      "loss": 4.282,
+      "step": 10880
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001230379339474637,
+      "loss": 2.4792,
+      "step": 10885
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00012299770706786274,
+      "loss": 1.8746,
+      "step": 10890
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001229574801882618,
+      "loss": 1.6147,
+      "step": 10895
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00012291725330866086,
+      "loss": 2.9426,
+      "step": 10900
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001228770264290599,
+      "loss": 4.616,
+      "step": 10905
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00012283679954945895,
+      "loss": 3.8428,
+      "step": 10910
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.000122796572669858,
+      "loss": 4.851,
+      "step": 10915
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00012275634579025707,
+      "loss": 3.6221,
+      "step": 10920
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001227161189106561,
+      "loss": 4.0699,
+      "step": 10925
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012267589203105516,
+      "loss": 4.5023,
+      "step": 10930
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012263566515145421,
+      "loss": 3.5705,
+      "step": 10935
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012259543827185327,
+      "loss": 2.7856,
+      "step": 10940
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001225552113922523,
+      "loss": 1.8233,
+      "step": 10945
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012251498451265136,
+      "loss": 2.3532,
+      "step": 10950
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012247475763305042,
+      "loss": 3.4496,
+      "step": 10955
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012243453075344945,
+      "loss": 3.6041,
+      "step": 10960
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001223943038738485,
+      "loss": 4.3311,
+      "step": 10965
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012235407699424757,
+      "loss": 4.1073,
+      "step": 10970
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012231385011464663,
+      "loss": 4.1563,
+      "step": 10975
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012227362323504566,
+      "loss": 2.2184,
+      "step": 10980
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012223339635544471,
+      "loss": 2.3406,
+      "step": 10985
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012219316947584377,
+      "loss": 1.6087,
+      "step": 10990
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001221529425962428,
+      "loss": 1.7286,
+      "step": 10995
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012211271571664186,
+      "loss": 0.5842,
+      "step": 11000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012207248883704092,
+      "loss": 4.0867,
+      "step": 11005
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012203226195743998,
+      "loss": 3.6726,
+      "step": 11010
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012199203507783902,
+      "loss": 4.5297,
+      "step": 11015
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012195180819823805,
+      "loss": 3.8204,
+      "step": 11020
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012191158131863711,
+      "loss": 3.8697,
+      "step": 11025
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012187135443903617,
+      "loss": 3.5434,
+      "step": 11030
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012183112755943522,
+      "loss": 2.4551,
+      "step": 11035
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012179090067983427,
+      "loss": 2.9122,
+      "step": 11040
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012175067380023333,
+      "loss": 2.9427,
+      "step": 11045
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012171044692063238,
+      "loss": 0.0149,
+      "step": 11050
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012167022004103141,
+      "loss": 3.976,
+      "step": 11055
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012162999316143047,
+      "loss": 4.292,
+      "step": 11060
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012158976628182952,
+      "loss": 3.5534,
+      "step": 11065
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012154953940222857,
+      "loss": 3.6272,
+      "step": 11070
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012150931252262763,
+      "loss": 3.5693,
+      "step": 11075
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012146908564302669,
+      "loss": 3.2917,
+      "step": 11080
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012142885876342573,
+      "loss": 3.7476,
+      "step": 11085
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012138863188382479,
+      "loss": 1.9143,
+      "step": 11090
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012134840500422382,
+      "loss": 1.1044,
+      "step": 11095
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012130817812462288,
+      "loss": 2.3956,
+      "step": 11100
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012126795124502192,
+      "loss": 4.0947,
+      "step": 11105
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012122772436542098,
+      "loss": 3.8414,
+      "step": 11110
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012118749748582004,
+      "loss": 3.4203,
+      "step": 11115
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012114727060621908,
+      "loss": 4.1299,
+      "step": 11120
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012110704372661814,
+      "loss": 1.8336,
+      "step": 11125
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012106681684701717,
+      "loss": 3.1511,
+      "step": 11130
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012102658996741623,
+      "loss": 1.5535,
+      "step": 11135
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012098636308781528,
+      "loss": 2.8977,
+      "step": 11140
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012094613620821433,
+      "loss": 1.0519,
+      "step": 11145
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012090590932861339,
+      "loss": 2.3733,
+      "step": 11150
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012086568244901244,
+      "loss": 6.4816,
+      "step": 11155
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001208254555694115,
+      "loss": 4.9588,
+      "step": 11160
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012078522868981055,
+      "loss": 3.2568,
+      "step": 11165
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012074500181020959,
+      "loss": 3.6896,
+      "step": 11170
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012070477493060863,
+      "loss": 4.2188,
+      "step": 11175
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012066454805100769,
+      "loss": 3.8143,
+      "step": 11180
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012062432117140675,
+      "loss": 2.433,
+      "step": 11185
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00012058409429180579,
+      "loss": 2.9424,
+      "step": 11190
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00012054386741220485,
+      "loss": 1.7503,
+      "step": 11195
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001205036405326039,
+      "loss": 1.471,
+      "step": 11200
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00012046341365300294,
+      "loss": 4.4582,
+      "step": 11205
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00012042318677340198,
+      "loss": 4.9609,
+      "step": 11210
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00012038295989380104,
+      "loss": 4.7346,
+      "step": 11215
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00012034273301420009,
+      "loss": 3.2718,
+      "step": 11220
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00012030250613459914,
+      "loss": 3.5242,
+      "step": 11225
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001202622792549982,
+      "loss": 3.4757,
+      "step": 11230
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00012022205237539725,
+      "loss": 1.8217,
+      "step": 11235
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001201818254957963,
+      "loss": 2.7196,
+      "step": 11240
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00012014159861619534,
+      "loss": 1.9612,
+      "step": 11245
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001201013717365944,
+      "loss": 1.22,
+      "step": 11250
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00012006114485699344,
+      "loss": 3.6163,
+      "step": 11255
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001200209179773925,
+      "loss": 3.748,
+      "step": 11260
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011998069109779156,
+      "loss": 3.8621,
+      "step": 11265
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001199404642181906,
+      "loss": 3.7291,
+      "step": 11270
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011990023733858966,
+      "loss": 3.5688,
+      "step": 11275
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011986001045898869,
+      "loss": 3.9327,
+      "step": 11280
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011981978357938775,
+      "loss": 2.0496,
+      "step": 11285
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001197795566997868,
+      "loss": 3.2875,
+      "step": 11290
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011973932982018585,
+      "loss": 3.1575,
+      "step": 11295
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011969910294058491,
+      "loss": 1.5065,
+      "step": 11300
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011965887606098395,
+      "loss": 5.9072,
+      "step": 11305
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011961864918138301,
+      "loss": 4.5477,
+      "step": 11310
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011957842230178207,
+      "loss": 4.2107,
+      "step": 11315
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001195381954221811,
+      "loss": 4.1152,
+      "step": 11320
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00011949796854258015,
+      "loss": 3.9164,
+      "step": 11325
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001194577416629792,
+      "loss": 3.1883,
+      "step": 11330
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00011941751478337826,
+      "loss": 3.7471,
+      "step": 11335
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00011937728790377731,
+      "loss": 2.6846,
+      "step": 11340
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00011933706102417637,
+      "loss": 2.8391,
+      "step": 11345
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001193048795204956,
+      "loss": 2.3922,
+      "step": 11350
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00011926465264089466,
+      "loss": 4.3721,
+      "step": 11355
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00011922442576129372,
+      "loss": 4.3652,
+      "step": 11360
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00011918419888169275,
+      "loss": 3.515,
+      "step": 11365
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00011914397200209179,
+      "loss": 4.5871,
+      "step": 11370
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00011910374512249085,
+      "loss": 3.9205,
+      "step": 11375
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00011906351824288991,
+      "loss": 3.4631,
+      "step": 11380
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00011902329136328895,
+      "loss": 3.6807,
+      "step": 11385
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00011898306448368801,
+      "loss": 2.6828,
+      "step": 11390
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00011894283760408707,
+      "loss": 2.0819,
+      "step": 11395
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001189026107244861,
+      "loss": 2.5945,
+      "step": 11400
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00011886238384488515,
+      "loss": 3.8611,
+      "step": 11405
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001188221569652842,
+      "loss": 4.0271,
+      "step": 11410
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00011878193008568326,
+      "loss": 5.5275,
+      "step": 11415
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001187417032060823,
+      "loss": 4.565,
+      "step": 11420
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00011870147632648136,
+      "loss": 3.5147,
+      "step": 11425
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00011866124944688042,
+      "loss": 3.1143,
+      "step": 11430
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00011862102256727947,
+      "loss": 1.3822,
+      "step": 11435
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001185807956876785,
+      "loss": 2.3071,
+      "step": 11440
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00011854056880807756,
+      "loss": 1.8316,
+      "step": 11445
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00011850034192847662,
+      "loss": 1.9333,
+      "step": 11450
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00011846011504887566,
+      "loss": 4.7771,
+      "step": 11455
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00011841988816927472,
+      "loss": 4.8422,
+      "step": 11460
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00011837966128967378,
+      "loss": 4.3121,
+      "step": 11465
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00011833943441007282,
+      "loss": 3.5387,
+      "step": 11470
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00011829920753047185,
+      "loss": 2.7474,
+      "step": 11475
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00011825898065087091,
+      "loss": 3.3034,
+      "step": 11480
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00011821875377126997,
+      "loss": 3.3162,
+      "step": 11485
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00011817852689166901,
+      "loss": 1.4119,
+      "step": 11490
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00011813830001206807,
+      "loss": 2.9623,
+      "step": 11495
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00011809807313246713,
+      "loss": 1.924,
+      "step": 11500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00011805784625286617,
+      "loss": 4.39,
+      "step": 11505
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00011801761937326523,
+      "loss": 4.1092,
+      "step": 11510
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00011797739249366426,
+      "loss": 4.3584,
+      "step": 11515
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00011793716561406332,
+      "loss": 4.9307,
+      "step": 11520
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00011789693873446237,
+      "loss": 3.7303,
+      "step": 11525
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00011785671185486143,
+      "loss": 2.9764,
+      "step": 11530
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00011781648497526048,
+      "loss": 2.5169,
+      "step": 11535
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00011777625809565953,
+      "loss": 2.986,
+      "step": 11540
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00011773603121605859,
+      "loss": 2.9573,
+      "step": 11545
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00011769580433645762,
+      "loss": 4.3488,
+      "step": 11550
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00011765557745685668,
+      "loss": 3.9223,
+      "step": 11555
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00011761535057725572,
+      "loss": 4.3891,
+      "step": 11560
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00011757512369765478,
+      "loss": 4.6504,
+      "step": 11565
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00011753489681805384,
+      "loss": 2.1885,
+      "step": 11570
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00011749466993845288,
+      "loss": 4.1299,
+      "step": 11575
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00011745444305885194,
+      "loss": 2.6887,
+      "step": 11580
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00011741421617925098,
+      "loss": 4.266,
+      "step": 11585
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00011737398929965002,
+      "loss": 2.6325,
+      "step": 11590
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00011733376242004907,
+      "loss": 2.8226,
+      "step": 11595
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00011729353554044813,
+      "loss": 0.0985,
+      "step": 11600
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00011725330866084718,
+      "loss": 4.3924,
+      "step": 11605
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00011721308178124624,
+      "loss": 4.9402,
+      "step": 11610
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001171728549016453,
+      "loss": 4.0807,
+      "step": 11615
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00011713262802204434,
+      "loss": 3.7963,
+      "step": 11620
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00011709240114244337,
+      "loss": 3.3341,
+      "step": 11625
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00011705217426284243,
+      "loss": 2.103,
+      "step": 11630
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00011701194738324149,
+      "loss": 2.1604,
+      "step": 11635
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00011697172050364053,
+      "loss": 3.5096,
+      "step": 11640
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00011693149362403959,
+      "loss": 4.5993,
+      "step": 11645
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00011689126674443865,
+      "loss": 3.1042,
+      "step": 11650
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00011685103986483769,
+      "loss": 4.4627,
+      "step": 11655
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00011681081298523675,
+      "loss": 4.0707,
+      "step": 11660
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00011677058610563578,
+      "loss": 4.1977,
+      "step": 11665
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00011673035922603484,
+      "loss": 3.0867,
+      "step": 11670
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00011669013234643388,
+      "loss": 3.8916,
+      "step": 11675
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00011664990546683294,
+      "loss": 2.4693,
+      "step": 11680
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.000116609678587232,
+      "loss": 3.8117,
+      "step": 11685
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00011656945170763104,
+      "loss": 2.3956,
+      "step": 11690
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001165292248280301,
+      "loss": 2.6879,
+      "step": 11695
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00011648899794842913,
+      "loss": 1.1392,
+      "step": 11700
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00011644877106882819,
+      "loss": 4.1566,
+      "step": 11705
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00011640854418922724,
+      "loss": 4.0279,
+      "step": 11710
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001163683173096263,
+      "loss": 3.9164,
+      "step": 11715
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00011632809043002535,
+      "loss": 2.289,
+      "step": 11720
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001162878635504244,
+      "loss": 4.4148,
+      "step": 11725
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00011624763667082346,
+      "loss": 2.4125,
+      "step": 11730
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00011620740979122251,
+      "loss": 1.2408,
+      "step": 11735
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00011616718291162155,
+      "loss": 1.2736,
+      "step": 11740
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00011612695603202059,
+      "loss": 1.0584,
+      "step": 11745
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00011608672915241965,
+      "loss": 2.8411,
+      "step": 11750
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00011604650227281871,
+      "loss": 3.5629,
+      "step": 11755
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00011600627539321775,
+      "loss": 4.7592,
+      "step": 11760
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00011596604851361681,
+      "loss": 4.0055,
+      "step": 11765
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00011592582163401587,
+      "loss": 4.5269,
+      "step": 11770
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001158855947544149,
+      "loss": 4.535,
+      "step": 11775
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00011584536787481394,
+      "loss": 1.9481,
+      "step": 11780
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.000115805140995213,
+      "loss": 2.4671,
+      "step": 11785
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00011576491411561206,
+      "loss": 3.4428,
+      "step": 11790
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001157246872360111,
+      "loss": 2.0589,
+      "step": 11795
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00011568446035641016,
+      "loss": 3.3335,
+      "step": 11800
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00011564423347680922,
+      "loss": 4.425,
+      "step": 11805
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00011560400659720827,
+      "loss": 4.8104,
+      "step": 11810
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001155637797176073,
+      "loss": 4.1982,
+      "step": 11815
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00011552355283800636,
+      "loss": 4.6156,
+      "step": 11820
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00011548332595840541,
+      "loss": 3.4463,
+      "step": 11825
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00011544309907880446,
+      "loss": 1.7727,
+      "step": 11830
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00011540287219920352,
+      "loss": 2.8218,
+      "step": 11835
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00011536264531960258,
+      "loss": 2.5171,
+      "step": 11840
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011532241844000162,
+      "loss": 3.8074,
+      "step": 11845
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011528219156040065,
+      "loss": 1.3458,
+      "step": 11850
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011524196468079971,
+      "loss": 4.7215,
+      "step": 11855
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011520173780119877,
+      "loss": 4.6002,
+      "step": 11860
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011516151092159781,
+      "loss": 4.7979,
+      "step": 11865
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011512128404199687,
+      "loss": 4.3117,
+      "step": 11870
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011508105716239593,
+      "loss": 2.6911,
+      "step": 11875
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011504083028279497,
+      "loss": 3.0803,
+      "step": 11880
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011500060340319403,
+      "loss": 2.7019,
+      "step": 11885
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011496037652359306,
+      "loss": 2.1551,
+      "step": 11890
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011492014964399211,
+      "loss": 2.693,
+      "step": 11895
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011487992276439117,
+      "loss": 0.6029,
+      "step": 11900
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011483969588479022,
+      "loss": 4.9437,
+      "step": 11905
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011479946900518927,
+      "loss": 4.216,
+      "step": 11910
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011475924212558833,
+      "loss": 3.3949,
+      "step": 11915
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011471901524598739,
+      "loss": 3.6592,
+      "step": 11920
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011467878836638643,
+      "loss": 2.7439,
+      "step": 11925
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011463856148678546,
+      "loss": 2.8549,
+      "step": 11930
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011459833460718452,
+      "loss": 2.6736,
+      "step": 11935
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011455810772758358,
+      "loss": 3.3242,
+      "step": 11940
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011451788084798262,
+      "loss": 1.7755,
+      "step": 11945
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011447765396838168,
+      "loss": 2.2492,
+      "step": 11950
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011443742708878074,
+      "loss": 4.5637,
+      "step": 11955
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011439720020917978,
+      "loss": 4.16,
+      "step": 11960
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011435697332957881,
+      "loss": 4.0373,
+      "step": 11965
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011431674644997787,
+      "loss": 3.9465,
+      "step": 11970
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011427651957037693,
+      "loss": 4.5678,
+      "step": 11975
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011423629269077598,
+      "loss": 4.6479,
+      "step": 11980
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011419606581117503,
+      "loss": 3.3561,
+      "step": 11985
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011415583893157409,
+      "loss": 2.7581,
+      "step": 11990
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011411561205197314,
+      "loss": 1.4079,
+      "step": 11995
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0001140753851723722,
+      "loss": 3.3122,
+      "step": 12000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011403515829277123,
+      "loss": 5.2654,
+      "step": 12005
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011399493141317028,
+      "loss": 4.2666,
+      "step": 12010
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011395470453356933,
+      "loss": 3.2514,
+      "step": 12015
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011391447765396839,
+      "loss": 3.2791,
+      "step": 12020
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011387425077436745,
+      "loss": 2.7815,
+      "step": 12025
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011383402389476649,
+      "loss": 3.698,
+      "step": 12030
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011379379701516555,
+      "loss": 3.4422,
+      "step": 12035
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011375357013556458,
+      "loss": 0.9724,
+      "step": 12040
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011371334325596364,
+      "loss": 1.8559,
+      "step": 12045
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011367311637636268,
+      "loss": 1.9751,
+      "step": 12050
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011363288949676174,
+      "loss": 3.3685,
+      "step": 12055
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0001135926626171608,
+      "loss": 5.3922,
+      "step": 12060
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011355243573755984,
+      "loss": 3.7494,
+      "step": 12065
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0001135122088579589,
+      "loss": 5.3334,
+      "step": 12070
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011347198197835796,
+      "loss": 3.8536,
+      "step": 12075
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011343175509875699,
+      "loss": 3.947,
+      "step": 12080
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011339152821915604,
+      "loss": 2.8934,
+      "step": 12085
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0001133513013395551,
+      "loss": 2.3866,
+      "step": 12090
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011331107445995415,
+      "loss": 1.5607,
+      "step": 12095
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0001132708475803532,
+      "loss": 1.2598,
+      "step": 12100
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011323062070075226,
+      "loss": 3.7342,
+      "step": 12105
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011319039382115131,
+      "loss": 4.8529,
+      "step": 12110
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011315016694155035,
+      "loss": 4.0742,
+      "step": 12115
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011310994006194939,
+      "loss": 5.4617,
+      "step": 12120
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011306971318234845,
+      "loss": 3.8857,
+      "step": 12125
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001130294863027475,
+      "loss": 2.9613,
+      "step": 12130
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011298925942314655,
+      "loss": 1.5732,
+      "step": 12135
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011294903254354561,
+      "loss": 1.9693,
+      "step": 12140
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011290880566394467,
+      "loss": 1.1998,
+      "step": 12145
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011286857878434371,
+      "loss": 3.2726,
+      "step": 12150
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011282835190474274,
+      "loss": 4.6418,
+      "step": 12155
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001127881250251418,
+      "loss": 4.0848,
+      "step": 12160
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011274789814554086,
+      "loss": 4.4574,
+      "step": 12165
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001127076712659399,
+      "loss": 4.4372,
+      "step": 12170
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011266744438633896,
+      "loss": 4.7805,
+      "step": 12175
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011262721750673802,
+      "loss": 2.6393,
+      "step": 12180
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011258699062713707,
+      "loss": 3.8619,
+      "step": 12185
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001125467637475361,
+      "loss": 3.5495,
+      "step": 12190
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011250653686793516,
+      "loss": 3.6609,
+      "step": 12195
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001124663099883342,
+      "loss": 2.3156,
+      "step": 12200
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011242608310873326,
+      "loss": 4.6609,
+      "step": 12205
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011238585622913232,
+      "loss": 4.3674,
+      "step": 12210
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011234562934953136,
+      "loss": 3.684,
+      "step": 12215
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011230540246993042,
+      "loss": 4.9451,
+      "step": 12220
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011226517559032948,
+      "loss": 2.7038,
+      "step": 12225
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011222494871072851,
+      "loss": 3.1783,
+      "step": 12230
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011218472183112755,
+      "loss": 2.1691,
+      "step": 12235
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011214449495152661,
+      "loss": 2.7589,
+      "step": 12240
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011210426807192567,
+      "loss": 1.9916,
+      "step": 12245
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011206404119232471,
+      "loss": 2.93,
+      "step": 12250
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011202381431272377,
+      "loss": 3.9955,
+      "step": 12255
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011198358743312283,
+      "loss": 4.1354,
+      "step": 12260
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011194336055352186,
+      "loss": 3.9764,
+      "step": 12265
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0001119031336739209,
+      "loss": 3.9035,
+      "step": 12270
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011186290679431996,
+      "loss": 3.8275,
+      "step": 12275
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011182267991471902,
+      "loss": 3.5661,
+      "step": 12280
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011178245303511807,
+      "loss": 3.2651,
+      "step": 12285
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011174222615551713,
+      "loss": 1.1449,
+      "step": 12290
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011170199927591618,
+      "loss": 1.265,
+      "step": 12295
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011166177239631523,
+      "loss": 0.3177,
+      "step": 12300
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011162154551671426,
+      "loss": 5.1195,
+      "step": 12305
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011158131863711332,
+      "loss": 4.2492,
+      "step": 12310
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011154109175751238,
+      "loss": 4.85,
+      "step": 12315
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011150086487791142,
+      "loss": 3.5096,
+      "step": 12320
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011146063799831048,
+      "loss": 3.4133,
+      "step": 12325
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011142041111870954,
+      "loss": 2.644,
+      "step": 12330
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011138018423910858,
+      "loss": 1.4073,
+      "step": 12335
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011133995735950761,
+      "loss": 2.6904,
+      "step": 12340
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011129973047990667,
+      "loss": 4.1987,
+      "step": 12345
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011125950360030573,
+      "loss": 4.0315,
+      "step": 12350
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011121927672070477,
+      "loss": 4.21,
+      "step": 12355
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011117904984110383,
+      "loss": 4.2699,
+      "step": 12360
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011113882296150289,
+      "loss": 5.3631,
+      "step": 12365
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011109859608190194,
+      "loss": 4.8795,
+      "step": 12370
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.000111058369202301,
+      "loss": 4.0154,
+      "step": 12375
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011101814232270003,
+      "loss": 4.0755,
+      "step": 12380
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011097791544309908,
+      "loss": 4.2021,
+      "step": 12385
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011093768856349813,
+      "loss": 2.2228,
+      "step": 12390
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011089746168389719,
+      "loss": 2.6391,
+      "step": 12395
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011085723480429624,
+      "loss": 2.3025,
+      "step": 12400
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011081700792469529,
+      "loss": 5.06,
+      "step": 12405
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011077678104509435,
+      "loss": 4.7277,
+      "step": 12410
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011073655416549338,
+      "loss": 4.8125,
+      "step": 12415
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011069632728589244,
+      "loss": 3.7311,
+      "step": 12420
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011065610040629148,
+      "loss": 3.3934,
+      "step": 12425
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011061587352669054,
+      "loss": 4.0301,
+      "step": 12430
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001105756466470896,
+      "loss": 3.5967,
+      "step": 12435
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011053541976748864,
+      "loss": 3.6605,
+      "step": 12440
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001104951928878877,
+      "loss": 2.1137,
+      "step": 12445
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011045496600828676,
+      "loss": 1.5456,
+      "step": 12450
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011041473912868579,
+      "loss": 4.5984,
+      "step": 12455
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011037451224908484,
+      "loss": 4.9756,
+      "step": 12460
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001103342853694839,
+      "loss": 4.725,
+      "step": 12465
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011029405848988295,
+      "loss": 5.0465,
+      "step": 12470
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.000110253831610282,
+      "loss": 3.6682,
+      "step": 12475
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011021360473068105,
+      "loss": 4.3359,
+      "step": 12480
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001101733778510801,
+      "loss": 2.6403,
+      "step": 12485
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011013315097147914,
+      "loss": 3.0468,
+      "step": 12490
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011009292409187819,
+      "loss": 3.0222,
+      "step": 12495
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00011005269721227725,
+      "loss": 0.6001,
+      "step": 12500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00011001247033267629,
+      "loss": 4.099,
+      "step": 12505
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00010997224345307535,
+      "loss": 3.7836,
+      "step": 12510
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00010993201657347441,
+      "loss": 4.1847,
+      "step": 12515
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00010989178969387345,
+      "loss": 4.2145,
+      "step": 12520
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00010985156281427251,
+      "loss": 3.6762,
+      "step": 12525
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00010981133593467154,
+      "loss": 3.2025,
+      "step": 12530
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001097711090550706,
+      "loss": 2.674,
+      "step": 12535
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00010973088217546965,
+      "loss": 2.7232,
+      "step": 12540
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001096906552958687,
+      "loss": 0.4914,
+      "step": 12545
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00010965042841626776,
+      "loss": 2.267,
+      "step": 12550
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001096102015366668,
+      "loss": 3.6879,
+      "step": 12555
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00010956997465706586,
+      "loss": 4.2727,
+      "step": 12560
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001095297477774649,
+      "loss": 3.6961,
+      "step": 12565
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00010948952089786395,
+      "loss": 2.9307,
+      "step": 12570
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.000109449294018263,
+      "loss": 2.4069,
+      "step": 12575
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00010940906713866206,
+      "loss": 2.6994,
+      "step": 12580
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00010936884025906112,
+      "loss": 1.9843,
+      "step": 12585
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00010932861337946016,
+      "loss": 2.8084,
+      "step": 12590
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00010928838649985922,
+      "loss": 2.6188,
+      "step": 12595
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00010924815962025828,
+      "loss": 0.5502,
+      "step": 12600
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00010920793274065731,
+      "loss": 5.0705,
+      "step": 12605
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00010916770586105635,
+      "loss": 3.4141,
+      "step": 12610
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00010912747898145541,
+      "loss": 3.7998,
+      "step": 12615
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00010908725210185447,
+      "loss": 3.3628,
+      "step": 12620
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00010904702522225351,
+      "loss": 3.8773,
+      "step": 12625
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010900679834265257,
+      "loss": 3.2954,
+      "step": 12630
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010896657146305163,
+      "loss": 3.2307,
+      "step": 12635
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010892634458345066,
+      "loss": 2.1269,
+      "step": 12640
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001088861177038497,
+      "loss": 2.2615,
+      "step": 12645
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010884589082424876,
+      "loss": 4.0116,
+      "step": 12650
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010880566394464782,
+      "loss": 4.5145,
+      "step": 12655
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010876543706504687,
+      "loss": 4.0908,
+      "step": 12660
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010872521018544593,
+      "loss": 3.6911,
+      "step": 12665
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010868498330584498,
+      "loss": 3.7748,
+      "step": 12670
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010864475642624403,
+      "loss": 3.6313,
+      "step": 12675
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010860452954664306,
+      "loss": 1.8294,
+      "step": 12680
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010856430266704212,
+      "loss": 3.2862,
+      "step": 12685
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010852407578744118,
+      "loss": 2.1133,
+      "step": 12690
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010848384890784022,
+      "loss": 2.6551,
+      "step": 12695
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010844362202823928,
+      "loss": 1.0988,
+      "step": 12700
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010840339514863834,
+      "loss": 4.7625,
+      "step": 12705
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010836316826903738,
+      "loss": 3.9336,
+      "step": 12710
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010832294138943641,
+      "loss": 4.7963,
+      "step": 12715
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010828271450983547,
+      "loss": 3.2798,
+      "step": 12720
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010824248763023453,
+      "loss": 2.7068,
+      "step": 12725
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010820226075063357,
+      "loss": 3.1443,
+      "step": 12730
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010816203387103263,
+      "loss": 3.0755,
+      "step": 12735
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010812180699143169,
+      "loss": 1.3621,
+      "step": 12740
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010808158011183074,
+      "loss": 2.0039,
+      "step": 12745
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001080413532322298,
+      "loss": 1.3085,
+      "step": 12750
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010800112635262882,
+      "loss": 5.2609,
+      "step": 12755
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010796089947302788,
+      "loss": 4.3895,
+      "step": 12760
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010792067259342693,
+      "loss": 3.3505,
+      "step": 12765
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010788044571382599,
+      "loss": 4.2628,
+      "step": 12770
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010784021883422504,
+      "loss": 5.0826,
+      "step": 12775
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010779999195462409,
+      "loss": 2.538,
+      "step": 12780
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010775976507502315,
+      "loss": 3.901,
+      "step": 12785
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010771953819542218,
+      "loss": 1.9004,
+      "step": 12790
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010767931131582124,
+      "loss": 0.5085,
+      "step": 12795
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010763908443622028,
+      "loss": 3.4105,
+      "step": 12800
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010759885755661934,
+      "loss": 3.8672,
+      "step": 12805
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010755863067701838,
+      "loss": 4.4018,
+      "step": 12810
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010751840379741744,
+      "loss": 3.9307,
+      "step": 12815
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001074781769178165,
+      "loss": 4.3627,
+      "step": 12820
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010743795003821555,
+      "loss": 3.4223,
+      "step": 12825
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010739772315861458,
+      "loss": 2.4649,
+      "step": 12830
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010735749627901363,
+      "loss": 1.3693,
+      "step": 12835
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010731726939941269,
+      "loss": 1.899,
+      "step": 12840
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010727704251981174,
+      "loss": 0.5113,
+      "step": 12845
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001072368156402108,
+      "loss": 2.3541,
+      "step": 12850
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010719658876060985,
+      "loss": 5.3012,
+      "step": 12855
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001071563618810089,
+      "loss": 3.8258,
+      "step": 12860
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010711613500140793,
+      "loss": 4.4615,
+      "step": 12865
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010707590812180699,
+      "loss": 4.375,
+      "step": 12870
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010703568124220605,
+      "loss": 3.2504,
+      "step": 12875
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010699545436260509,
+      "loss": 2.7794,
+      "step": 12880
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010695522748300415,
+      "loss": 2.613,
+      "step": 12885
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010691500060340321,
+      "loss": 2.0133,
+      "step": 12890
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010687477372380225,
+      "loss": 3.0668,
+      "step": 12895
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010683454684420131,
+      "loss": 2.141,
+      "step": 12900
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010679431996460034,
+      "loss": 3.9652,
+      "step": 12905
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001067540930849994,
+      "loss": 5.0807,
+      "step": 12910
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010671386620539844,
+      "loss": 4.0959,
+      "step": 12915
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001066736393257975,
+      "loss": 2.3016,
+      "step": 12920
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010663341244619656,
+      "loss": 3.6994,
+      "step": 12925
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001065931855665956,
+      "loss": 3.4188,
+      "step": 12930
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010655295868699466,
+      "loss": 2.1822,
+      "step": 12935
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001065127318073937,
+      "loss": 2.6569,
+      "step": 12940
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010647250492779275,
+      "loss": 2.9767,
+      "step": 12945
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001064322780481918,
+      "loss": 0.6595,
+      "step": 12950
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010639205116859086,
+      "loss": 4.4453,
+      "step": 12955
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010635182428898991,
+      "loss": 4.6098,
+      "step": 12960
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010631159740938896,
+      "loss": 4.4469,
+      "step": 12965
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010627137052978802,
+      "loss": 4.1836,
+      "step": 12970
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010623114365018708,
+      "loss": 3.3398,
+      "step": 12975
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001061909167705861,
+      "loss": 3.8299,
+      "step": 12980
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010615068989098515,
+      "loss": 2.2403,
+      "step": 12985
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010611046301138421,
+      "loss": 1.9197,
+      "step": 12990
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010607023613178327,
+      "loss": 1.7764,
+      "step": 12995
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010603000925218231,
+      "loss": 3.4165,
+      "step": 13000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010598978237258137,
+      "loss": 5.0324,
+      "step": 13005
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010594955549298043,
+      "loss": 4.4018,
+      "step": 13010
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010590932861337946,
+      "loss": 4.1213,
+      "step": 13015
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001058691017337785,
+      "loss": 4.3871,
+      "step": 13020
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010582887485417756,
+      "loss": 2.7148,
+      "step": 13025
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010578864797457662,
+      "loss": 2.8373,
+      "step": 13030
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010574842109497567,
+      "loss": 2.7298,
+      "step": 13035
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010570819421537472,
+      "loss": 0.7858,
+      "step": 13040
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010566796733577378,
+      "loss": 1.4459,
+      "step": 13045
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010562774045617283,
+      "loss": 2.2835,
+      "step": 13050
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010558751357657186,
+      "loss": 4.6979,
+      "step": 13055
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010554728669697092,
+      "loss": 3.1915,
+      "step": 13060
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010550705981736997,
+      "loss": 2.701,
+      "step": 13065
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010546683293776902,
+      "loss": 3.3332,
+      "step": 13070
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010542660605816808,
+      "loss": 3.1688,
+      "step": 13075
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010538637917856714,
+      "loss": 2.2415,
+      "step": 13080
+    },
+    {
+      "epoch": 1.0,
+      "eval_exact_match": 19.315068493150687,
+      "eval_f1": 23.273953174088238,
+      "eval_loss": 3.1703224182128906,
+      "eval_runtime": 132.4755,
+      "eval_samples_per_second": 11.021,
+      "eval_steps_per_second": 11.021,
+      "step": 13084
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010534615229896618,
+      "loss": 5.7792,
+      "step": 13085
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010530592541936521,
+      "loss": 3.9447,
+      "step": 13090
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010526569853976427,
+      "loss": 3.901,
+      "step": 13095
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010522547166016333,
+      "loss": 4.2629,
+      "step": 13100
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010518524478056237,
+      "loss": 3.0552,
+      "step": 13105
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010514501790096143,
+      "loss": 2.5984,
+      "step": 13110
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010510479102136048,
+      "loss": 2.634,
+      "step": 13115
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010506456414175953,
+      "loss": 2.7914,
+      "step": 13120
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010502433726215859,
+      "loss": 2.0628,
+      "step": 13125
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010498411038255762,
+      "loss": 3.1447,
+      "step": 13130
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010494388350295667,
+      "loss": 2.7328,
+      "step": 13135
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010490365662335573,
+      "loss": 3.76,
+      "step": 13140
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010486342974375478,
+      "loss": 4.6025,
+      "step": 13145
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010482320286415383,
+      "loss": 3.5472,
+      "step": 13150
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010478297598455289,
+      "loss": 3.857,
+      "step": 13155
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010474274910495195,
+      "loss": 2.5723,
+      "step": 13160
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010470252222535098,
+      "loss": 2.8614,
+      "step": 13165
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010466229534575002,
+      "loss": 2.782,
+      "step": 13170
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010462206846614908,
+      "loss": 2.5527,
+      "step": 13175
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010458184158654814,
+      "loss": 2.8946,
+      "step": 13180
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010454161470694718,
+      "loss": 2.8931,
+      "step": 13185
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010450138782734624,
+      "loss": 4.4236,
+      "step": 13190
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001044611609477453,
+      "loss": 4.7928,
+      "step": 13195
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010442093406814434,
+      "loss": 3.9273,
+      "step": 13200
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010438070718854338,
+      "loss": 3.9992,
+      "step": 13205
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010434048030894243,
+      "loss": 4.2393,
+      "step": 13210
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010430025342934149,
+      "loss": 3.4548,
+      "step": 13215
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010426002654974054,
+      "loss": 3.0828,
+      "step": 13220
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001042197996701396,
+      "loss": 2.7957,
+      "step": 13225
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010417957279053865,
+      "loss": 2.4393,
+      "step": 13230
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001041393459109377,
+      "loss": 3.4124,
+      "step": 13235
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010409911903133673,
+      "loss": 4.693,
+      "step": 13240
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010405889215173579,
+      "loss": 3.4697,
+      "step": 13245
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010401866527213485,
+      "loss": 2.767,
+      "step": 13250
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010397843839253389,
+      "loss": 3.817,
+      "step": 13255
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010393821151293295,
+      "loss": 1.5805,
+      "step": 13260
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.000103897984633332,
+      "loss": 2.659,
+      "step": 13265
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010385775775373105,
+      "loss": 1.4231,
+      "step": 13270
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010381753087413011,
+      "loss": 1.9977,
+      "step": 13275
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010377730399452914,
+      "loss": 1.0283,
+      "step": 13280
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001037370771149282,
+      "loss": 4.241,
+      "step": 13285
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00010369685023532724,
+      "loss": 3.7126,
+      "step": 13290
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001036566233557263,
+      "loss": 3.5291,
+      "step": 13295
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00010361639647612536,
+      "loss": 4.2404,
+      "step": 13300
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001035761695965244,
+      "loss": 3.9266,
+      "step": 13305
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00010353594271692346,
+      "loss": 3.1577,
+      "step": 13310
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001034957158373225,
+      "loss": 4.3291,
+      "step": 13315
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00010345548895772155,
+      "loss": 2.2583,
+      "step": 13320
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001034152620781206,
+      "loss": 2.4731,
+      "step": 13325
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00010337503519851966,
+      "loss": 2.9766,
+      "step": 13330
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00010333480831891871,
+      "loss": 3.4053,
+      "step": 13335
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00010329458143931776,
+      "loss": 5.1129,
+      "step": 13340
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00010325435455971682,
+      "loss": 4.6836,
+      "step": 13345
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00010321412768011587,
+      "loss": 3.6654,
+      "step": 13350
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001031739008005149,
+      "loss": 4.1396,
+      "step": 13355
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00010313367392091395,
+      "loss": 3.5745,
+      "step": 13360
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00010309344704131301,
+      "loss": 2.6486,
+      "step": 13365
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00010305322016171207,
+      "loss": 2.9922,
+      "step": 13370
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00010301299328211111,
+      "loss": 2.7838,
+      "step": 13375
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00010297276640251017,
+      "loss": 1.5658,
+      "step": 13380
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00010293253952290923,
+      "loss": 1.1819,
+      "step": 13385
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00010289231264330826,
+      "loss": 3.4799,
+      "step": 13390
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001028520857637073,
+      "loss": 4.5635,
+      "step": 13395
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00010281185888410636,
+      "loss": 3.1668,
+      "step": 13400
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00010277163200450542,
+      "loss": 3.4604,
+      "step": 13405
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00010273140512490447,
+      "loss": 2.7437,
+      "step": 13410
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00010269117824530352,
+      "loss": 1.6988,
+      "step": 13415
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00010265095136570257,
+      "loss": 3.8053,
+      "step": 13420
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00010261072448610163,
+      "loss": 1.8165,
+      "step": 13425
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00010257049760650066,
+      "loss": 1.5206,
+      "step": 13430
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00010253027072689972,
+      "loss": 1.837,
+      "step": 13435
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00010249004384729876,
+      "loss": 3.6209,
+      "step": 13440
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00010244981696769782,
+      "loss": 5.1863,
+      "step": 13445
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00010240959008809688,
+      "loss": 3.5703,
+      "step": 13450
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00010236936320849592,
+      "loss": 4.4045,
+      "step": 13455
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00010232913632889498,
+      "loss": 2.1611,
+      "step": 13460
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00010228890944929401,
+      "loss": 1.9873,
+      "step": 13465
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00010224868256969307,
+      "loss": 2.744,
+      "step": 13470
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00010220845569009211,
+      "loss": 3.7924,
+      "step": 13475
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00010216822881049117,
+      "loss": 1.564,
+      "step": 13480
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00010212800193089023,
+      "loss": 1.5397,
+      "step": 13485
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00010208777505128927,
+      "loss": 4.3014,
+      "step": 13490
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00010204754817168833,
+      "loss": 4.2459,
+      "step": 13495
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00010200732129208739,
+      "loss": 3.9824,
+      "step": 13500
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00010196709441248642,
+      "loss": 3.8135,
+      "step": 13505
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00010192686753288547,
+      "loss": 2.5058,
+      "step": 13510
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00010188664065328453,
+      "loss": 3.7534,
+      "step": 13515
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00010184641377368358,
+      "loss": 3.8482,
+      "step": 13520
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00010180618689408263,
+      "loss": 2.4387,
+      "step": 13525
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00010176596001448169,
+      "loss": 1.9385,
+      "step": 13530
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00010172573313488074,
+      "loss": 2.6273,
+      "step": 13535
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00010168550625527978,
+      "loss": 4.2504,
+      "step": 13540
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00010164527937567882,
+      "loss": 4.3891,
+      "step": 13545
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00010160505249607788,
+      "loss": 4.7129,
+      "step": 13550
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00010156482561647694,
+      "loss": 4.8672,
+      "step": 13555
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00010152459873687598,
+      "loss": 4.241,
+      "step": 13560
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00010148437185727504,
+      "loss": 3.9742,
+      "step": 13565
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0001014441449776741,
+      "loss": 2.0136,
+      "step": 13570
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00010140391809807314,
+      "loss": 1.6322,
+      "step": 13575
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00010136369121847217,
+      "loss": 1.4756,
+      "step": 13580
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00010132346433887123,
+      "loss": 4.3438,
+      "step": 13585
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00010128323745927029,
+      "loss": 4.408,
+      "step": 13590
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00010124301057966934,
+      "loss": 3.9469,
+      "step": 13595
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0001012027837000684,
+      "loss": 3.242,
+      "step": 13600
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00010116255682046745,
+      "loss": 3.3846,
+      "step": 13605
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0001011223299408665,
+      "loss": 3.5865,
+      "step": 13610
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00010108210306126553,
+      "loss": 0.9825,
+      "step": 13615
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00010104187618166459,
+      "loss": 1.6402,
+      "step": 13620
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00010100164930206364,
+      "loss": 1.2693,
+      "step": 13625
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00010096142242246269,
+      "loss": 2.1932,
+      "step": 13630
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00010092119554286175,
+      "loss": 3.0698,
+      "step": 13635
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0001008809686632608,
+      "loss": 4.5355,
+      "step": 13640
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00010084074178365985,
+      "loss": 4.377,
+      "step": 13645
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00010080051490405891,
+      "loss": 2.7748,
+      "step": 13650
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00010076028802445794,
+      "loss": 2.7549,
+      "step": 13655
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.000100720061144857,
+      "loss": 2.8492,
+      "step": 13660
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00010067983426525604,
+      "loss": 2.1549,
+      "step": 13665
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0001006396073856551,
+      "loss": 4.6738,
+      "step": 13670
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00010059938050605416,
+      "loss": 3.3374,
+      "step": 13675
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0001005591536264532,
+      "loss": 1.8652,
+      "step": 13680
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00010051892674685226,
+      "loss": 1.9717,
+      "step": 13685
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00010047869986725129,
+      "loss": 4.8148,
+      "step": 13690
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00010043847298765035,
+      "loss": 2.9995,
+      "step": 13695
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0001003982461080494,
+      "loss": 3.6135,
+      "step": 13700
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00010035801922844845,
+      "loss": 3.8224,
+      "step": 13705
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00010031779234884751,
+      "loss": 3.3087,
+      "step": 13710
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00010027756546924656,
+      "loss": 3.1735,
+      "step": 13715
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00010023733858964562,
+      "loss": 2.4742,
+      "step": 13720
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00010019711171004466,
+      "loss": 1.4694,
+      "step": 13725
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00010015688483044369,
+      "loss": 0.4333,
+      "step": 13730
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00010011665795084275,
+      "loss": 1.2572,
+      "step": 13735
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00010007643107124181,
+      "loss": 3.6918,
+      "step": 13740
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00010003620419164085,
+      "loss": 3.8943,
+      "step": 13745
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.999597731203991e-05,
+      "loss": 4.2775,
+      "step": 13750
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.995575043243897e-05,
+      "loss": 3.6002,
+      "step": 13755
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.991552355283801e-05,
+      "loss": 2.8794,
+      "step": 13760
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.987529667323706e-05,
+      "loss": 3.1105,
+      "step": 13765
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.983506979363612e-05,
+      "loss": 4.7389,
+      "step": 13770
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.979484291403516e-05,
+      "loss": 2.5333,
+      "step": 13775
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.97546160344342e-05,
+      "loss": 2.6911,
+      "step": 13780
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.971438915483326e-05,
+      "loss": 0.6666,
+      "step": 13785
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.967416227523232e-05,
+      "loss": 4.4268,
+      "step": 13790
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.963393539563137e-05,
+      "loss": 4.1672,
+      "step": 13795
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.959370851603041e-05,
+      "loss": 3.8434,
+      "step": 13800
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.955348163642947e-05,
+      "loss": 4.1248,
+      "step": 13805
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.951325475682853e-05,
+      "loss": 3.2066,
+      "step": 13810
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.947302787722756e-05,
+      "loss": 3.5237,
+      "step": 13815
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.943280099762662e-05,
+      "loss": 3.5643,
+      "step": 13820
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.939257411802568e-05,
+      "loss": 3.0314,
+      "step": 13825
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.935234723842472e-05,
+      "loss": 1.1349,
+      "step": 13830
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.931212035882377e-05,
+      "loss": 3.0678,
+      "step": 13835
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.927189347922282e-05,
+      "loss": 3.6559,
+      "step": 13840
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.923166659962188e-05,
+      "loss": 4.8559,
+      "step": 13845
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.919143972002091e-05,
+      "loss": 3.8783,
+      "step": 13850
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.915121284041997e-05,
+      "loss": 4.1764,
+      "step": 13855
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.911098596081903e-05,
+      "loss": 3.7188,
+      "step": 13860
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.907075908121807e-05,
+      "loss": 2.8204,
+      "step": 13865
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.903053220161712e-05,
+      "loss": 3.4747,
+      "step": 13870
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.899030532201618e-05,
+      "loss": 2.0576,
+      "step": 13875
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.895007844241522e-05,
+      "loss": 1.8002,
+      "step": 13880
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.890985156281428e-05,
+      "loss": 1.943,
+      "step": 13885
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.886962468321332e-05,
+      "loss": 5.2752,
+      "step": 13890
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.882939780361238e-05,
+      "loss": 3.9357,
+      "step": 13895
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.878917092401143e-05,
+      "loss": 3.7773,
+      "step": 13900
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.874894404441049e-05,
+      "loss": 5.2211,
+      "step": 13905
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.870871716480953e-05,
+      "loss": 3.1906,
+      "step": 13910
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.866849028520858e-05,
+      "loss": 2.1916,
+      "step": 13915
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.862826340560763e-05,
+      "loss": 3.045,
+      "step": 13920
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.858803652600668e-05,
+      "loss": 3.3197,
+      "step": 13925
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.854780964640574e-05,
+      "loss": 1.3265,
+      "step": 13930
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.850758276680478e-05,
+      "loss": 2.4261,
+      "step": 13935
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.846735588720384e-05,
+      "loss": 4.1789,
+      "step": 13940
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.842712900760288e-05,
+      "loss": 4.1479,
+      "step": 13945
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.838690212800193e-05,
+      "loss": 3.1523,
+      "step": 13950
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.834667524840099e-05,
+      "loss": 3.0566,
+      "step": 13955
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.830644836880005e-05,
+      "loss": 3.6089,
+      "step": 13960
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.826622148919909e-05,
+      "loss": 2.8464,
+      "step": 13965
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.822599460959813e-05,
+      "loss": 1.619,
+      "step": 13970
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.818576772999719e-05,
+      "loss": 2.2897,
+      "step": 13975
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.814554085039625e-05,
+      "loss": 1.4078,
+      "step": 13980
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.810531397079528e-05,
+      "loss": 2.4326,
+      "step": 13985
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.806508709119434e-05,
+      "loss": 4.1578,
+      "step": 13990
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.80248602115934e-05,
+      "loss": 4.7398,
+      "step": 13995
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.798463333199244e-05,
+      "loss": 3.6618,
+      "step": 14000
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.794440645239149e-05,
+      "loss": 3.7291,
+      "step": 14005
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.790417957279055e-05,
+      "loss": 2.9765,
+      "step": 14010
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.78639526931896e-05,
+      "loss": 3.6363,
+      "step": 14015
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.782372581358864e-05,
+      "loss": 1.3663,
+      "step": 14020
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.77834989339877e-05,
+      "loss": 3.0463,
+      "step": 14025
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.774327205438675e-05,
+      "loss": 5.0727,
+      "step": 14030
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.77030451747858e-05,
+      "loss": 1.6521,
+      "step": 14035
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.766281829518484e-05,
+      "loss": 4.0453,
+      "step": 14040
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.76225914155839e-05,
+      "loss": 4.0281,
+      "step": 14045
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.758236453598294e-05,
+      "loss": 4.8402,
+      "step": 14050
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.7542137656382e-05,
+      "loss": 4.2115,
+      "step": 14055
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.750191077678105e-05,
+      "loss": 2.9648,
+      "step": 14060
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.74616838971801e-05,
+      "loss": 2.7151,
+      "step": 14065
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.742145701757915e-05,
+      "loss": 2.5117,
+      "step": 14070
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.73812301379782e-05,
+      "loss": 1.9299,
+      "step": 14075
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.734100325837725e-05,
+      "loss": 2.7252,
+      "step": 14080
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.73007763787763e-05,
+      "loss": 2.3402,
+      "step": 14085
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.726054949917536e-05,
+      "loss": 4.2039,
+      "step": 14090
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.72203226195744e-05,
+      "loss": 4.5305,
+      "step": 14095
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.718009573997346e-05,
+      "loss": 3.88,
+      "step": 14100
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.71398688603725e-05,
+      "loss": 3.9152,
+      "step": 14105
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.709964198077156e-05,
+      "loss": 2.9659,
+      "step": 14110
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.70594151011706e-05,
+      "loss": 1.5826,
+      "step": 14115
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.701918822156965e-05,
+      "loss": 2.4249,
+      "step": 14120
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.697896134196871e-05,
+      "loss": 1.3763,
+      "step": 14125
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.693873446236777e-05,
+      "loss": 3.5167,
+      "step": 14130
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.689850758276681e-05,
+      "loss": 1.3236,
+      "step": 14135
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.685828070316586e-05,
+      "loss": 3.5751,
+      "step": 14140
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.681805382356492e-05,
+      "loss": 5.0203,
+      "step": 14145
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.677782694396396e-05,
+      "loss": 4.1728,
+      "step": 14150
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.6737600064363e-05,
+      "loss": 3.4594,
+      "step": 14155
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.669737318476206e-05,
+      "loss": 3.3553,
+      "step": 14160
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.665714630516112e-05,
+      "loss": 2.2479,
+      "step": 14165
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.661691942556017e-05,
+      "loss": 1.877,
+      "step": 14170
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.657669254595921e-05,
+      "loss": 2.6314,
+      "step": 14175
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.653646566635827e-05,
+      "loss": 2.0204,
+      "step": 14180
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.649623878675731e-05,
+      "loss": 0.9242,
+      "step": 14185
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.645601190715636e-05,
+      "loss": 4.19,
+      "step": 14190
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.641578502755542e-05,
+      "loss": 3.0912,
+      "step": 14195
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.637555814795447e-05,
+      "loss": 4.9994,
+      "step": 14200
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.633533126835352e-05,
+      "loss": 3.509,
+      "step": 14205
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.629510438875256e-05,
+      "loss": 2.7194,
+      "step": 14210
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.625487750915162e-05,
+      "loss": 2.4299,
+      "step": 14215
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.621465062955067e-05,
+      "loss": 2.4398,
+      "step": 14220
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.617442374994971e-05,
+      "loss": 2.8874,
+      "step": 14225
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.613419687034877e-05,
+      "loss": 2.0132,
+      "step": 14230
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.609396999074783e-05,
+      "loss": 2.243,
+      "step": 14235
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.605374311114687e-05,
+      "loss": 4.0598,
+      "step": 14240
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.601351623154592e-05,
+      "loss": 5.1547,
+      "step": 14245
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.597328935194498e-05,
+      "loss": 3.9782,
+      "step": 14250
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.593306247234402e-05,
+      "loss": 2.596,
+      "step": 14255
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.589283559274308e-05,
+      "loss": 1.8018,
+      "step": 14260
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.585260871314212e-05,
+      "loss": 3.1381,
+      "step": 14265
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.581238183354118e-05,
+      "loss": 3.5054,
+      "step": 14270
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.577215495394023e-05,
+      "loss": 2.9537,
+      "step": 14275
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.573192807433928e-05,
+      "loss": 1.9567,
+      "step": 14280
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.569170119473833e-05,
+      "loss": 2.3284,
+      "step": 14285
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.565147431513737e-05,
+      "loss": 3.6228,
+      "step": 14290
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.561124743553643e-05,
+      "loss": 3.4064,
+      "step": 14295
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.557102055593548e-05,
+      "loss": 4.2512,
+      "step": 14300
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.553079367633454e-05,
+      "loss": 3.4361,
+      "step": 14305
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.549056679673358e-05,
+      "loss": 3.2916,
+      "step": 14310
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.545033991713264e-05,
+      "loss": 4.3375,
+      "step": 14315
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.541011303753168e-05,
+      "loss": 1.8709,
+      "step": 14320
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.536988615793073e-05,
+      "loss": 1.8777,
+      "step": 14325
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.532965927832979e-05,
+      "loss": 3.9809,
+      "step": 14330
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.528943239872884e-05,
+      "loss": 4.1262,
+      "step": 14335
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.524920551912788e-05,
+      "loss": 4.7285,
+      "step": 14340
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.520897863952693e-05,
+      "loss": 4.2479,
+      "step": 14345
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.516875175992599e-05,
+      "loss": 3.2966,
+      "step": 14350
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.512852488032504e-05,
+      "loss": 3.9391,
+      "step": 14355
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.508829800072408e-05,
+      "loss": 3.4237,
+      "step": 14360
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.504807112112314e-05,
+      "loss": 2.8419,
+      "step": 14365
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.50078442415222e-05,
+      "loss": 3.4014,
+      "step": 14370
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.496761736192123e-05,
+      "loss": 2.3402,
+      "step": 14375
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.492739048232029e-05,
+      "loss": 2.5134,
+      "step": 14380
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.488716360271935e-05,
+      "loss": 2.5162,
+      "step": 14385
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.484693672311839e-05,
+      "loss": 4.0307,
+      "step": 14390
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.480670984351743e-05,
+      "loss": 4.8123,
+      "step": 14395
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.476648296391649e-05,
+      "loss": 3.1288,
+      "step": 14400
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.472625608431555e-05,
+      "loss": 3.1115,
+      "step": 14405
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.46860292047146e-05,
+      "loss": 2.6729,
+      "step": 14410
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.464580232511364e-05,
+      "loss": 1.9552,
+      "step": 14415
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.46055754455127e-05,
+      "loss": 3.342,
+      "step": 14420
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.456534856591174e-05,
+      "loss": 0.694,
+      "step": 14425
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.45251216863108e-05,
+      "loss": 1.77,
+      "step": 14430
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.448489480670985e-05,
+      "loss": 1.6854,
+      "step": 14435
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.44446679271089e-05,
+      "loss": 4.6209,
+      "step": 14440
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.440444104750795e-05,
+      "loss": 4.9201,
+      "step": 14445
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.4364214167907e-05,
+      "loss": 4.5273,
+      "step": 14450
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.432398728830605e-05,
+      "loss": 3.6521,
+      "step": 14455
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.42837604087051e-05,
+      "loss": 3.382,
+      "step": 14460
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.424353352910416e-05,
+      "loss": 2.8238,
+      "step": 14465
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.42033066495032e-05,
+      "loss": 3.1938,
+      "step": 14470
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.416307976990226e-05,
+      "loss": 1.7498,
+      "step": 14475
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.41228528903013e-05,
+      "loss": 2.284,
+      "step": 14480
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.408262601070036e-05,
+      "loss": 2.6607,
+      "step": 14485
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.40423991310994e-05,
+      "loss": 3.5379,
+      "step": 14490
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.400217225149845e-05,
+      "loss": 4.8506,
+      "step": 14495
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.396194537189751e-05,
+      "loss": 4.3186,
+      "step": 14500
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.392171849229657e-05,
+      "loss": 4.0508,
+      "step": 14505
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.38814916126956e-05,
+      "loss": 3.2298,
+      "step": 14510
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.384126473309466e-05,
+      "loss": 2.3462,
+      "step": 14515
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.380103785349371e-05,
+      "loss": 1.9607,
+      "step": 14520
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.376081097389276e-05,
+      "loss": 1.1034,
+      "step": 14525
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.37205840942918e-05,
+      "loss": 2.063,
+      "step": 14530
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.368035721469086e-05,
+      "loss": 2.0346,
+      "step": 14535
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.364013033508992e-05,
+      "loss": 4.0316,
+      "step": 14540
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.359990345548895e-05,
+      "loss": 4.7727,
+      "step": 14545
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.355967657588801e-05,
+      "loss": 4.61,
+      "step": 14550
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.351944969628707e-05,
+      "loss": 3.921,
+      "step": 14555
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.347922281668611e-05,
+      "loss": 3.5495,
+      "step": 14560
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.343899593708516e-05,
+      "loss": 3.8674,
+      "step": 14565
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.339876905748422e-05,
+      "loss": 2.3444,
+      "step": 14570
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.335854217788327e-05,
+      "loss": 2.3662,
+      "step": 14575
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.331831529828232e-05,
+      "loss": 1.6259,
+      "step": 14580
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.327808841868136e-05,
+      "loss": 2.9731,
+      "step": 14585
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.323786153908042e-05,
+      "loss": 5.4543,
+      "step": 14590
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.319763465947947e-05,
+      "loss": 3.7518,
+      "step": 14595
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.315740777987851e-05,
+      "loss": 2.8011,
+      "step": 14600
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.311718090027757e-05,
+      "loss": 2.6506,
+      "step": 14605
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.307695402067663e-05,
+      "loss": 2.9476,
+      "step": 14610
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.303672714107567e-05,
+      "loss": 2.7561,
+      "step": 14615
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.299650026147472e-05,
+      "loss": 1.8628,
+      "step": 14620
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.295627338187378e-05,
+      "loss": 3.3484,
+      "step": 14625
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.291604650227282e-05,
+      "loss": 1.8501,
+      "step": 14630
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.287581962267188e-05,
+      "loss": 1.1852,
+      "step": 14635
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.283559274307092e-05,
+      "loss": 3.9387,
+      "step": 14640
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.279536586346997e-05,
+      "loss": 5.0062,
+      "step": 14645
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.275513898386903e-05,
+      "loss": 3.9559,
+      "step": 14650
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.271491210426808e-05,
+      "loss": 3.1137,
+      "step": 14655
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.267468522466713e-05,
+      "loss": 2.2882,
+      "step": 14660
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.263445834506617e-05,
+      "loss": 3.7979,
+      "step": 14665
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.259423146546523e-05,
+      "loss": 1.49,
+      "step": 14670
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.255400458586428e-05,
+      "loss": 2.9072,
+      "step": 14675
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.251377770626332e-05,
+      "loss": 0.0377,
+      "step": 14680
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.247355082666238e-05,
+      "loss": 0.8162,
+      "step": 14685
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.243332394706144e-05,
+      "loss": 4.1197,
+      "step": 14690
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.239309706746048e-05,
+      "loss": 3.5349,
+      "step": 14695
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.235287018785953e-05,
+      "loss": 3.0641,
+      "step": 14700
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.231264330825858e-05,
+      "loss": 3.3049,
+      "step": 14705
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.227241642865764e-05,
+      "loss": 2.9953,
+      "step": 14710
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.223218954905667e-05,
+      "loss": 4.6273,
+      "step": 14715
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.219196266945573e-05,
+      "loss": 3.2683,
+      "step": 14720
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.215173578985479e-05,
+      "loss": 1.9925,
+      "step": 14725
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.211150891025384e-05,
+      "loss": 4.6559,
+      "step": 14730
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.207128203065288e-05,
+      "loss": 2.2516,
+      "step": 14735
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.203105515105194e-05,
+      "loss": 5.6625,
+      "step": 14740
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.1990828271451e-05,
+      "loss": 3.9943,
+      "step": 14745
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.195060139185003e-05,
+      "loss": 3.9695,
+      "step": 14750
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.191037451224909e-05,
+      "loss": 4.2107,
+      "step": 14755
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.187014763264814e-05,
+      "loss": 3.3329,
+      "step": 14760
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.182992075304719e-05,
+      "loss": 3.2871,
+      "step": 14765
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.178969387344623e-05,
+      "loss": 4.1098,
+      "step": 14770
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.174946699384529e-05,
+      "loss": 1.0645,
+      "step": 14775
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.170924011424435e-05,
+      "loss": 0.74,
+      "step": 14780
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.16690132346434e-05,
+      "loss": 4.3225,
+      "step": 14785
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.162878635504244e-05,
+      "loss": 4.441,
+      "step": 14790
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.15885594754415e-05,
+      "loss": 3.9297,
+      "step": 14795
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.154833259584054e-05,
+      "loss": 3.2412,
+      "step": 14800
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.15081057162396e-05,
+      "loss": 3.4386,
+      "step": 14805
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.146787883663865e-05,
+      "loss": 3.839,
+      "step": 14810
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.142765195703769e-05,
+      "loss": 4.0252,
+      "step": 14815
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.138742507743675e-05,
+      "loss": 2.962,
+      "step": 14820
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.13471981978358e-05,
+      "loss": 2.9706,
+      "step": 14825
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.130697131823485e-05,
+      "loss": 3.4539,
+      "step": 14830
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.12667444386339e-05,
+      "loss": 3.3337,
+      "step": 14835
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.122651755903295e-05,
+      "loss": 4.6289,
+      "step": 14840
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.1186290679432e-05,
+      "loss": 4.4992,
+      "step": 14845
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.114606379983104e-05,
+      "loss": 4.495,
+      "step": 14850
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.11058369202301e-05,
+      "loss": 3.5725,
+      "step": 14855
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.106561004062916e-05,
+      "loss": 2.8839,
+      "step": 14860
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.10253831610282e-05,
+      "loss": 2.2224,
+      "step": 14865
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.098515628142725e-05,
+      "loss": 2.9777,
+      "step": 14870
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.094492940182631e-05,
+      "loss": 0.4112,
+      "step": 14875
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.090470252222537e-05,
+      "loss": 3.0038,
+      "step": 14880
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.08644756426244e-05,
+      "loss": 2.8895,
+      "step": 14885
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.082424876302346e-05,
+      "loss": 4.4019,
+      "step": 14890
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.078402188342251e-05,
+      "loss": 4.5066,
+      "step": 14895
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.074379500382156e-05,
+      "loss": 4.3509,
+      "step": 14900
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.07035681242206e-05,
+      "loss": 3.8162,
+      "step": 14905
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.066334124461966e-05,
+      "loss": 2.8266,
+      "step": 14910
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.062311436501872e-05,
+      "loss": 1.7993,
+      "step": 14915
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.058288748541775e-05,
+      "loss": 3.0204,
+      "step": 14920
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.054266060581681e-05,
+      "loss": 2.9019,
+      "step": 14925
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.050243372621587e-05,
+      "loss": 1.8711,
+      "step": 14930
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.046220684661491e-05,
+      "loss": 0.6589,
+      "step": 14935
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.042197996701396e-05,
+      "loss": 4.4053,
+      "step": 14940
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.038175308741301e-05,
+      "loss": 4.7037,
+      "step": 14945
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.034152620781206e-05,
+      "loss": 3.6162,
+      "step": 14950
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.030129932821112e-05,
+      "loss": 3.1366,
+      "step": 14955
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.026107244861016e-05,
+      "loss": 4.0396,
+      "step": 14960
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.022084556900922e-05,
+      "loss": 2.0449,
+      "step": 14965
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.018061868940827e-05,
+      "loss": 2.2303,
+      "step": 14970
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.014039180980732e-05,
+      "loss": 2.6239,
+      "step": 14975
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.010016493020637e-05,
+      "loss": 2.1086,
+      "step": 14980
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.005993805060541e-05,
+      "loss": 5.5885,
+      "step": 14985
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.001971117100447e-05,
+      "loss": 4.2605,
+      "step": 14990
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.997948429140352e-05,
+      "loss": 4.6322,
+      "step": 14995
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.993925741180257e-05,
+      "loss": 3.768,
+      "step": 15000
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.989903053220162e-05,
+      "loss": 3.6373,
+      "step": 15005
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.985880365260068e-05,
+      "loss": 3.0281,
+      "step": 15010
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.981857677299972e-05,
+      "loss": 3.4137,
+      "step": 15015
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.977834989339877e-05,
+      "loss": 2.8749,
+      "step": 15020
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.973812301379782e-05,
+      "loss": 0.759,
+      "step": 15025
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.969789613419688e-05,
+      "loss": 2.8243,
+      "step": 15030
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.965766925459593e-05,
+      "loss": 2.4736,
+      "step": 15035
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.961744237499497e-05,
+      "loss": 4.4713,
+      "step": 15040
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.957721549539403e-05,
+      "loss": 4.3701,
+      "step": 15045
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.953698861579309e-05,
+      "loss": 4.4148,
+      "step": 15050
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.949676173619212e-05,
+      "loss": 3.6678,
+      "step": 15055
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.945653485659118e-05,
+      "loss": 3.7158,
+      "step": 15060
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.941630797699024e-05,
+      "loss": 3.458,
+      "step": 15065
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.937608109738928e-05,
+      "loss": 3.1514,
+      "step": 15070
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.933585421778833e-05,
+      "loss": 1.927,
+      "step": 15075
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.929562733818738e-05,
+      "loss": 2.2447,
+      "step": 15080
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.925540045858644e-05,
+      "loss": 2.924,
+      "step": 15085
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.921517357898547e-05,
+      "loss": 3.8148,
+      "step": 15090
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.917494669938453e-05,
+      "loss": 3.6268,
+      "step": 15095
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.913471981978359e-05,
+      "loss": 3.8186,
+      "step": 15100
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.909449294018263e-05,
+      "loss": 4.7416,
+      "step": 15105
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.905426606058168e-05,
+      "loss": 3.0722,
+      "step": 15110
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.901403918098074e-05,
+      "loss": 2.4213,
+      "step": 15115
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.897381230137978e-05,
+      "loss": 1.8222,
+      "step": 15120
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.893358542177884e-05,
+      "loss": 2.871,
+      "step": 15125
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.889335854217789e-05,
+      "loss": 1.8803,
+      "step": 15130
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.885313166257694e-05,
+      "loss": 2.8122,
+      "step": 15135
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.881290478297599e-05,
+      "loss": 3.7818,
+      "step": 15140
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.877267790337503e-05,
+      "loss": 3.7229,
+      "step": 15145
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.873245102377409e-05,
+      "loss": 4.0561,
+      "step": 15150
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.869222414417314e-05,
+      "loss": 3.7717,
+      "step": 15155
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.86519972645722e-05,
+      "loss": 3.7115,
+      "step": 15160
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.861177038497124e-05,
+      "loss": 3.7303,
+      "step": 15165
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.85715435053703e-05,
+      "loss": 2.5779,
+      "step": 15170
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.853131662576934e-05,
+      "loss": 3.3491,
+      "step": 15175
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.84910897461684e-05,
+      "loss": 1.5736,
+      "step": 15180
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.845086286656744e-05,
+      "loss": 2.1361,
+      "step": 15185
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.841063598696649e-05,
+      "loss": 4.5369,
+      "step": 15190
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.837040910736555e-05,
+      "loss": 4.5656,
+      "step": 15195
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.83301822277646e-05,
+      "loss": 2.8299,
+      "step": 15200
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.828995534816365e-05,
+      "loss": 4.0764,
+      "step": 15205
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.82497284685627e-05,
+      "loss": 3.6823,
+      "step": 15210
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.820950158896175e-05,
+      "loss": 2.6001,
+      "step": 15215
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.81692747093608e-05,
+      "loss": 2.5453,
+      "step": 15220
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.812904782975984e-05,
+      "loss": 3.1553,
+      "step": 15225
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.80888209501589e-05,
+      "loss": 2.2298,
+      "step": 15230
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.804859407055796e-05,
+      "loss": 1.4315,
+      "step": 15235
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.8008367190957e-05,
+      "loss": 4.1703,
+      "step": 15240
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.796814031135605e-05,
+      "loss": 4.1615,
+      "step": 15245
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.792791343175511e-05,
+      "loss": 4.0225,
+      "step": 15250
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.788768655215415e-05,
+      "loss": 3.3277,
+      "step": 15255
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.78474596725532e-05,
+      "loss": 4.7938,
+      "step": 15260
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.780723279295225e-05,
+      "loss": 3.795,
+      "step": 15265
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.776700591335131e-05,
+      "loss": 3.2113,
+      "step": 15270
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.772677903375036e-05,
+      "loss": 1.447,
+      "step": 15275
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.76865521541494e-05,
+      "loss": 1.6464,
+      "step": 15280
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.764632527454846e-05,
+      "loss": 2.8649,
+      "step": 15285
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.76060983949475e-05,
+      "loss": 4.4762,
+      "step": 15290
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.756587151534655e-05,
+      "loss": 4.0934,
+      "step": 15295
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.752564463574561e-05,
+      "loss": 4.1266,
+      "step": 15300
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.748541775614467e-05,
+      "loss": 3.9455,
+      "step": 15305
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.744519087654371e-05,
+      "loss": 2.9204,
+      "step": 15310
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.740496399694276e-05,
+      "loss": 3.201,
+      "step": 15315
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.736473711734181e-05,
+      "loss": 3.3204,
+      "step": 15320
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.732451023774086e-05,
+      "loss": 1.0861,
+      "step": 15325
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.728428335813992e-05,
+      "loss": 2.0387,
+      "step": 15330
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.724405647853896e-05,
+      "loss": 2.5072,
+      "step": 15335
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.720382959893802e-05,
+      "loss": 3.6967,
+      "step": 15340
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.716360271933706e-05,
+      "loss": 4.2148,
+      "step": 15345
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.712337583973612e-05,
+      "loss": 2.8939,
+      "step": 15350
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.708314896013517e-05,
+      "loss": 3.6441,
+      "step": 15355
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.704292208053421e-05,
+      "loss": 2.4613,
+      "step": 15360
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.700269520093327e-05,
+      "loss": 1.0514,
+      "step": 15365
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.696246832133231e-05,
+      "loss": 4.2469,
+      "step": 15370
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.692224144173137e-05,
+      "loss": 3.054,
+      "step": 15375
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.688201456213042e-05,
+      "loss": 1.4101,
+      "step": 15380
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.684178768252948e-05,
+      "loss": 2.7099,
+      "step": 15385
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.680156080292852e-05,
+      "loss": 4.5326,
+      "step": 15390
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.676133392332757e-05,
+      "loss": 3.8502,
+      "step": 15395
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.672110704372662e-05,
+      "loss": 3.8027,
+      "step": 15400
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.668088016412568e-05,
+      "loss": 3.708,
+      "step": 15405
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.664065328452471e-05,
+      "loss": 3.4665,
+      "step": 15410
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.660042640492377e-05,
+      "loss": 2.2829,
+      "step": 15415
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.656019952532283e-05,
+      "loss": 2.9721,
+      "step": 15420
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.651997264572187e-05,
+      "loss": 2.7123,
+      "step": 15425
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.647974576612092e-05,
+      "loss": 1.2022,
+      "step": 15430
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.643951888651998e-05,
+      "loss": 1.9478,
+      "step": 15435
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.639929200691904e-05,
+      "loss": 4.2762,
+      "step": 15440
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.635906512731807e-05,
+      "loss": 3.9773,
+      "step": 15445
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.631883824771712e-05,
+      "loss": 4.3887,
+      "step": 15450
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.627861136811618e-05,
+      "loss": 4.1759,
+      "step": 15455
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.623838448851523e-05,
+      "loss": 2.006,
+      "step": 15460
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.619815760891427e-05,
+      "loss": 3.5844,
+      "step": 15465
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.615793072931333e-05,
+      "loss": 4.083,
+      "step": 15470
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.611770384971239e-05,
+      "loss": 2.3228,
+      "step": 15475
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.607747697011143e-05,
+      "loss": 3.9055,
+      "step": 15480
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.603725009051048e-05,
+      "loss": 1.9998,
+      "step": 15485
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.599702321090954e-05,
+      "loss": 4.3775,
+      "step": 15490
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.595679633130858e-05,
+      "loss": 5.6619,
+      "step": 15495
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.591656945170764e-05,
+      "loss": 3.8799,
+      "step": 15500
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.587634257210668e-05,
+      "loss": 4.2123,
+      "step": 15505
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.583611569250574e-05,
+      "loss": 3.1585,
+      "step": 15510
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.579588881290479e-05,
+      "loss": 3.1555,
+      "step": 15515
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.575566193330383e-05,
+      "loss": 2.6523,
+      "step": 15520
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.571543505370289e-05,
+      "loss": 1.6222,
+      "step": 15525
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.567520817410193e-05,
+      "loss": 1.5623,
+      "step": 15530
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.563498129450099e-05,
+      "loss": 2.1904,
+      "step": 15535
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.559475441490004e-05,
+      "loss": 4.3115,
+      "step": 15540
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.55545275352991e-05,
+      "loss": 4.4436,
+      "step": 15545
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.551430065569814e-05,
+      "loss": 4.1313,
+      "step": 15550
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.54740737760972e-05,
+      "loss": 3.7244,
+      "step": 15555
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.543384689649624e-05,
+      "loss": 1.7071,
+      "step": 15560
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.539362001689529e-05,
+      "loss": 2.7745,
+      "step": 15565
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.536143851321453e-05,
+      "loss": 3.1961,
+      "step": 15570
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.532121163361358e-05,
+      "loss": 1.68,
+      "step": 15575
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.528098475401264e-05,
+      "loss": 2.2349,
+      "step": 15580
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.524075787441168e-05,
+      "loss": 3.1614,
+      "step": 15585
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.520053099481074e-05,
+      "loss": 5.0854,
+      "step": 15590
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.516030411520979e-05,
+      "loss": 4.314,
+      "step": 15595
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.512007723560884e-05,
+      "loss": 3.1324,
+      "step": 15600
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.507985035600789e-05,
+      "loss": 4.0867,
+      "step": 15605
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.503962347640693e-05,
+      "loss": 2.0064,
+      "step": 15610
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.499939659680599e-05,
+      "loss": 2.5377,
+      "step": 15615
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.495916971720505e-05,
+      "loss": 3.2949,
+      "step": 15620
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.491894283760408e-05,
+      "loss": 1.9116,
+      "step": 15625
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.487871595800314e-05,
+      "loss": 1.14,
+      "step": 15630
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.48384890784022e-05,
+      "loss": 4.8432,
+      "step": 15635
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.479826219880124e-05,
+      "loss": 4.5416,
+      "step": 15640
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.475803531920029e-05,
+      "loss": 4.0746,
+      "step": 15645
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.471780843959934e-05,
+      "loss": 3.4386,
+      "step": 15650
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.46775815599984e-05,
+      "loss": 3.0038,
+      "step": 15655
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.463735468039743e-05,
+      "loss": 3.2941,
+      "step": 15660
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.459712780079649e-05,
+      "loss": 3.2194,
+      "step": 15665
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.455690092119555e-05,
+      "loss": 1.8783,
+      "step": 15670
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.45166740415946e-05,
+      "loss": 2.5379,
+      "step": 15675
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.447644716199364e-05,
+      "loss": 2.8854,
+      "step": 15680
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.44362202823927e-05,
+      "loss": 2.5177,
+      "step": 15685
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.439599340279176e-05,
+      "loss": 4.4383,
+      "step": 15690
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.43557665231908e-05,
+      "loss": 3.7588,
+      "step": 15695
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.431553964358985e-05,
+      "loss": 4.1584,
+      "step": 15700
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.42753127639889e-05,
+      "loss": 4.5521,
+      "step": 15705
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.423508588438795e-05,
+      "loss": 3.7143,
+      "step": 15710
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.419485900478701e-05,
+      "loss": 3.5352,
+      "step": 15715
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.415463212518605e-05,
+      "loss": 2.5749,
+      "step": 15720
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.411440524558511e-05,
+      "loss": 2.019,
+      "step": 15725
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.407417836598415e-05,
+      "loss": 1.2621,
+      "step": 15730
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.40339514863832e-05,
+      "loss": 1.5193,
+      "step": 15735
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.399372460678226e-05,
+      "loss": 4.5795,
+      "step": 15740
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.39534977271813e-05,
+      "loss": 4.4549,
+      "step": 15745
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.391327084758036e-05,
+      "loss": 3.8432,
+      "step": 15750
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.38730439679794e-05,
+      "loss": 4.1367,
+      "step": 15755
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.383281708837846e-05,
+      "loss": 2.8366,
+      "step": 15760
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.379259020877751e-05,
+      "loss": 3.5475,
+      "step": 15765
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.375236332917657e-05,
+      "loss": 3.802,
+      "step": 15770
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.371213644957561e-05,
+      "loss": 2.1158,
+      "step": 15775
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.367190956997466e-05,
+      "loss": 2.9005,
+      "step": 15780
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.363168269037371e-05,
+      "loss": 4.0268,
+      "step": 15785
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.359145581077277e-05,
+      "loss": 3.5148,
+      "step": 15790
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.35512289311718e-05,
+      "loss": 3.9832,
+      "step": 15795
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.351100205157086e-05,
+      "loss": 4.1332,
+      "step": 15800
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.347077517196992e-05,
+      "loss": 2.8575,
+      "step": 15805
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.343054829236896e-05,
+      "loss": 2.4339,
+      "step": 15810
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.339032141276801e-05,
+      "loss": 2.5007,
+      "step": 15815
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.335009453316707e-05,
+      "loss": 2.438,
+      "step": 15820
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.330986765356613e-05,
+      "loss": 2.3291,
+      "step": 15825
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.326964077396516e-05,
+      "loss": 1.1762,
+      "step": 15830
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.322941389436422e-05,
+      "loss": 1.7898,
+      "step": 15835
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.318918701476327e-05,
+      "loss": 4.6023,
+      "step": 15840
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.314896013516232e-05,
+      "loss": 4.2715,
+      "step": 15845
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.310873325556136e-05,
+      "loss": 4.5984,
+      "step": 15850
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.306850637596042e-05,
+      "loss": 3.9041,
+      "step": 15855
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.302827949635948e-05,
+      "loss": 3.0089,
+      "step": 15860
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.298805261675852e-05,
+      "loss": 3.0388,
+      "step": 15865
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.294782573715757e-05,
+      "loss": 3.5033,
+      "step": 15870
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.290759885755663e-05,
+      "loss": 3.0061,
+      "step": 15875
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.286737197795567e-05,
+      "loss": 2.2896,
+      "step": 15880
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.282714509835472e-05,
+      "loss": 2.4662,
+      "step": 15885
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.278691821875377e-05,
+      "loss": 4.1061,
+      "step": 15890
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.274669133915283e-05,
+      "loss": 4.6795,
+      "step": 15895
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.270646445955188e-05,
+      "loss": 4.7134,
+      "step": 15900
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.266623757995092e-05,
+      "loss": 3.7715,
+      "step": 15905
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.262601070034998e-05,
+      "loss": 3.1307,
+      "step": 15910
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.258578382074903e-05,
+      "loss": 1.9352,
+      "step": 15915
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.254555694114808e-05,
+      "loss": 1.595,
+      "step": 15920
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.250533006154713e-05,
+      "loss": 1.327,
+      "step": 15925
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.246510318194617e-05,
+      "loss": 1.5466,
+      "step": 15930
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.242487630234523e-05,
+      "loss": 2.0442,
+      "step": 15935
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.238464942274429e-05,
+      "loss": 5.4271,
+      "step": 15940
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.234442254314333e-05,
+      "loss": 4.067,
+      "step": 15945
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.230419566354238e-05,
+      "loss": 3.5479,
+      "step": 15950
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.226396878394144e-05,
+      "loss": 3.7012,
+      "step": 15955
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.222374190434048e-05,
+      "loss": 3.7094,
+      "step": 15960
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.218351502473953e-05,
+      "loss": 3.0054,
+      "step": 15965
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.214328814513858e-05,
+      "loss": 2.5629,
+      "step": 15970
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.210306126553764e-05,
+      "loss": 1.7412,
+      "step": 15975
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.206283438593669e-05,
+      "loss": 2.4695,
+      "step": 15980
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.202260750633573e-05,
+      "loss": 3.7052,
+      "step": 15985
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.198238062673479e-05,
+      "loss": 4.3586,
+      "step": 15990
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.194215374713385e-05,
+      "loss": 4.7576,
+      "step": 15995
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.190192686753288e-05,
+      "loss": 4.1148,
+      "step": 16000
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.186169998793194e-05,
+      "loss": 4.3701,
+      "step": 16005
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.1821473108331e-05,
+      "loss": 2.6996,
+      "step": 16010
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.178124622873004e-05,
+      "loss": 3.8242,
+      "step": 16015
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.174101934912909e-05,
+      "loss": 1.7617,
+      "step": 16020
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.170079246952814e-05,
+      "loss": 3.2437,
+      "step": 16025
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.16605655899272e-05,
+      "loss": 2.0556,
+      "step": 16030
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.162033871032623e-05,
+      "loss": 1.7406,
+      "step": 16035
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.158011183072529e-05,
+      "loss": 3.6297,
+      "step": 16040
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.153988495112435e-05,
+      "loss": 3.8234,
+      "step": 16045
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.14996580715234e-05,
+      "loss": 3.1287,
+      "step": 16050
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.145943119192244e-05,
+      "loss": 3.4059,
+      "step": 16055
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.14192043123215e-05,
+      "loss": 3.0411,
+      "step": 16060
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.137897743272056e-05,
+      "loss": 1.7471,
+      "step": 16065
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.13387505531196e-05,
+      "loss": 2.7685,
+      "step": 16070
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.129852367351864e-05,
+      "loss": 1.8943,
+      "step": 16075
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.12582967939177e-05,
+      "loss": 2.1345,
+      "step": 16080
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.121806991431675e-05,
+      "loss": 0.4706,
+      "step": 16085
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.11778430347158e-05,
+      "loss": 4.9555,
+      "step": 16090
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.113761615511485e-05,
+      "loss": 3.7021,
+      "step": 16095
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.10973892755139e-05,
+      "loss": 3.6622,
+      "step": 16100
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.105716239591295e-05,
+      "loss": 3.2216,
+      "step": 16105
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.1016935516312e-05,
+      "loss": 3.7122,
+      "step": 16110
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.097670863671106e-05,
+      "loss": 3.2248,
+      "step": 16115
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.09364817571101e-05,
+      "loss": 2.1737,
+      "step": 16120
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.089625487750916e-05,
+      "loss": 3.0768,
+      "step": 16125
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.08560279979082e-05,
+      "loss": 1.7508,
+      "step": 16130
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.081580111830725e-05,
+      "loss": 1.544,
+      "step": 16135
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.077557423870631e-05,
+      "loss": 4.6045,
+      "step": 16140
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.073534735910537e-05,
+      "loss": 4.5873,
+      "step": 16145
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.069512047950441e-05,
+      "loss": 4.3631,
+      "step": 16150
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.065489359990345e-05,
+      "loss": 4.2217,
+      "step": 16155
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.061466672030251e-05,
+      "loss": 2.9564,
+      "step": 16160
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.057443984070157e-05,
+      "loss": 3.6113,
+      "step": 16165
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.05342129611006e-05,
+      "loss": 2.8933,
+      "step": 16170
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.049398608149966e-05,
+      "loss": 3.0604,
+      "step": 16175
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.045375920189872e-05,
+      "loss": 3.6262,
+      "step": 16180
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.041353232229776e-05,
+      "loss": 3.0539,
+      "step": 16185
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.037330544269681e-05,
+      "loss": 4.0107,
+      "step": 16190
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.033307856309587e-05,
+      "loss": 4.325,
+      "step": 16195
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.029285168349492e-05,
+      "loss": 3.9152,
+      "step": 16200
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.025262480389396e-05,
+      "loss": 2.3825,
+      "step": 16205
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.021239792429301e-05,
+      "loss": 3.0158,
+      "step": 16210
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.017217104469207e-05,
+      "loss": 3.0349,
+      "step": 16215
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.013194416509112e-05,
+      "loss": 2.9871,
+      "step": 16220
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.009171728549016e-05,
+      "loss": 3.3361,
+      "step": 16225
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.005149040588922e-05,
+      "loss": 2.4099,
+      "step": 16230
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.001126352628826e-05,
+      "loss": 4.984,
+      "step": 16235
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.997103664668732e-05,
+      "loss": 4.5197,
+      "step": 16240
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.993080976708637e-05,
+      "loss": 4.6932,
+      "step": 16245
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.989058288748543e-05,
+      "loss": 3.5482,
+      "step": 16250
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.985035600788447e-05,
+      "loss": 4.3166,
+      "step": 16255
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.981012912828353e-05,
+      "loss": 3.2842,
+      "step": 16260
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.976990224868257e-05,
+      "loss": 3.2132,
+      "step": 16265
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.972967536908162e-05,
+      "loss": 3.3531,
+      "step": 16270
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.968944848948068e-05,
+      "loss": 3.0101,
+      "step": 16275
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.964922160987972e-05,
+      "loss": 3.3757,
+      "step": 16280
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.960899473027878e-05,
+      "loss": 2.4891,
+      "step": 16285
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.956876785067782e-05,
+      "loss": 4.3678,
+      "step": 16290
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.952854097107688e-05,
+      "loss": 4.4967,
+      "step": 16295
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.948831409147593e-05,
+      "loss": 4.7766,
+      "step": 16300
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.944808721187497e-05,
+      "loss": 4.4363,
+      "step": 16305
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.940786033227403e-05,
+      "loss": 3.5326,
+      "step": 16310
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.936763345267309e-05,
+      "loss": 2.0882,
+      "step": 16315
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.932740657307213e-05,
+      "loss": 2.1091,
+      "step": 16320
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.928717969347118e-05,
+      "loss": 2.316,
+      "step": 16325
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.924695281387024e-05,
+      "loss": 3.159,
+      "step": 16330
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.92067259342693e-05,
+      "loss": 1.6889,
+      "step": 16335
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.916649905466833e-05,
+      "loss": 3.8652,
+      "step": 16340
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.912627217506738e-05,
+      "loss": 4.3504,
+      "step": 16345
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.908604529546644e-05,
+      "loss": 4.0221,
+      "step": 16350
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.904581841586549e-05,
+      "loss": 4.374,
+      "step": 16355
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.900559153626453e-05,
+      "loss": 4.5057,
+      "step": 16360
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.896536465666359e-05,
+      "loss": 2.6971,
+      "step": 16365
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.892513777706265e-05,
+      "loss": 1.9066,
+      "step": 16370
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.888491089746168e-05,
+      "loss": 2.0673,
+      "step": 16375
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.884468401786074e-05,
+      "loss": 2.7272,
+      "step": 16380
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.88044571382598e-05,
+      "loss": 3.6661,
+      "step": 16385
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.876423025865884e-05,
+      "loss": 4.982,
+      "step": 16390
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.872400337905788e-05,
+      "loss": 3.9939,
+      "step": 16395
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.868377649945694e-05,
+      "loss": 4.1607,
+      "step": 16400
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.864354961985599e-05,
+      "loss": 3.9865,
+      "step": 16405
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.860332274025505e-05,
+      "loss": 4.0355,
+      "step": 16410
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.856309586065409e-05,
+      "loss": 3.2419,
+      "step": 16415
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.852286898105315e-05,
+      "loss": 2.5692,
+      "step": 16420
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.84826421014522e-05,
+      "loss": 3.0755,
+      "step": 16425
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.844241522185124e-05,
+      "loss": 1.1237,
+      "step": 16430
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.84021883422503e-05,
+      "loss": 2.8103,
+      "step": 16435
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.836196146264934e-05,
+      "loss": 4.2614,
+      "step": 16440
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.83217345830484e-05,
+      "loss": 4.7865,
+      "step": 16445
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.828150770344744e-05,
+      "loss": 4.392,
+      "step": 16450
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.82412808238465e-05,
+      "loss": 2.8524,
+      "step": 16455
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.820105394424555e-05,
+      "loss": 2.9417,
+      "step": 16460
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.81608270646446e-05,
+      "loss": 3.8519,
+      "step": 16465
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.812060018504365e-05,
+      "loss": 3.798,
+      "step": 16470
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.80803733054427e-05,
+      "loss": 3.2817,
+      "step": 16475
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.804014642584175e-05,
+      "loss": 0.6551,
+      "step": 16480
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.799991954624081e-05,
+      "loss": 0.9354,
+      "step": 16485
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.795969266663986e-05,
+      "loss": 4.2775,
+      "step": 16490
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.79194657870389e-05,
+      "loss": 4.2949,
+      "step": 16495
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.787923890743796e-05,
+      "loss": 4.0965,
+      "step": 16500
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.7839012027837e-05,
+      "loss": 3.2883,
+      "step": 16505
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.779878514823605e-05,
+      "loss": 3.9234,
+      "step": 16510
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.77585582686351e-05,
+      "loss": 2.299,
+      "step": 16515
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.771833138903416e-05,
+      "loss": 1.2933,
+      "step": 16520
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.767810450943321e-05,
+      "loss": 0.9545,
+      "step": 16525
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.763787762983225e-05,
+      "loss": 2.7195,
+      "step": 16530
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.759765075023131e-05,
+      "loss": 3.0816,
+      "step": 16535
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.755742387063036e-05,
+      "loss": 3.9545,
+      "step": 16540
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.75171969910294e-05,
+      "loss": 4.1322,
+      "step": 16545
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.747697011142846e-05,
+      "loss": 4.1234,
+      "step": 16550
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.743674323182752e-05,
+      "loss": 3.8441,
+      "step": 16555
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.739651635222656e-05,
+      "loss": 4.11,
+      "step": 16560
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.735628947262561e-05,
+      "loss": 1.9523,
+      "step": 16565
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.731606259302467e-05,
+      "loss": 2.1999,
+      "step": 16570
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.727583571342371e-05,
+      "loss": 2.8916,
+      "step": 16575
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.723560883382276e-05,
+      "loss": 0.9832,
+      "step": 16580
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.719538195422181e-05,
+      "loss": 2.7123,
+      "step": 16585
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.715515507462087e-05,
+      "loss": 4.9367,
+      "step": 16590
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.711492819501992e-05,
+      "loss": 3.8705,
+      "step": 16595
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.707470131541896e-05,
+      "loss": 3.7537,
+      "step": 16600
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.703447443581802e-05,
+      "loss": 3.0968,
+      "step": 16605
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.699424755621706e-05,
+      "loss": 2.827,
+      "step": 16610
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.695402067661612e-05,
+      "loss": 3.0993,
+      "step": 16615
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.691379379701517e-05,
+      "loss": 0.7201,
+      "step": 16620
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.687356691741423e-05,
+      "loss": 2.1911,
+      "step": 16625
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.683334003781327e-05,
+      "loss": 0.3554,
+      "step": 16630
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.679311315821233e-05,
+      "loss": 1.0917,
+      "step": 16635
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.675288627861137e-05,
+      "loss": 4.0449,
+      "step": 16640
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.671265939901042e-05,
+      "loss": 4.4773,
+      "step": 16645
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.667243251940948e-05,
+      "loss": 4.5842,
+      "step": 16650
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.663220563980852e-05,
+      "loss": 4.2728,
+      "step": 16655
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.659197876020758e-05,
+      "loss": 3.2278,
+      "step": 16660
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.655175188060662e-05,
+      "loss": 3.1791,
+      "step": 16665
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.651152500100568e-05,
+      "loss": 2.4688,
+      "step": 16670
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.647129812140473e-05,
+      "loss": 3.6795,
+      "step": 16675
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.643107124180377e-05,
+      "loss": 1.2943,
+      "step": 16680
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.639084436220283e-05,
+      "loss": 3.7789,
+      "step": 16685
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.635061748260189e-05,
+      "loss": 5.2229,
+      "step": 16690
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.631039060300092e-05,
+      "loss": 4.4244,
+      "step": 16695
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.627016372339998e-05,
+      "loss": 2.974,
+      "step": 16700
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.622993684379903e-05,
+      "loss": 3.6898,
+      "step": 16705
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.618970996419808e-05,
+      "loss": 2.8105,
+      "step": 16710
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.614948308459712e-05,
+      "loss": 1.7367,
+      "step": 16715
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.610925620499618e-05,
+      "loss": 2.6505,
+      "step": 16720
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.606902932539524e-05,
+      "loss": 1.5005,
+      "step": 16725
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.602880244579427e-05,
+      "loss": 1.0236,
+      "step": 16730
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.598857556619333e-05,
+      "loss": 3.0022,
+      "step": 16735
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.594834868659239e-05,
+      "loss": 4.4209,
+      "step": 16740
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.590812180699143e-05,
+      "loss": 4.0451,
+      "step": 16745
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.586789492739048e-05,
+      "loss": 4.0252,
+      "step": 16750
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.582766804778954e-05,
+      "loss": 2.3807,
+      "step": 16755
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.57874411681886e-05,
+      "loss": 3.1335,
+      "step": 16760
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.574721428858764e-05,
+      "loss": 3.6875,
+      "step": 16765
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.570698740898668e-05,
+      "loss": 2.3758,
+      "step": 16770
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.566676052938574e-05,
+      "loss": 1.6693,
+      "step": 16775
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.562653364978479e-05,
+      "loss": 2.396,
+      "step": 16780
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.558630677018384e-05,
+      "loss": 1.7757,
+      "step": 16785
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.554607989058289e-05,
+      "loss": 4.6365,
+      "step": 16790
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.550585301098195e-05,
+      "loss": 3.7377,
+      "step": 16795
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.546562613138099e-05,
+      "loss": 3.5774,
+      "step": 16800
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.542539925178004e-05,
+      "loss": 3.6809,
+      "step": 16805
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.53851723721791e-05,
+      "loss": 3.406,
+      "step": 16810
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.534494549257814e-05,
+      "loss": 2.5013,
+      "step": 16815
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.53047186129772e-05,
+      "loss": 3.9091,
+      "step": 16820
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.526449173337624e-05,
+      "loss": 3.2679,
+      "step": 16825
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.52242648537753e-05,
+      "loss": 3.0857,
+      "step": 16830
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.518403797417435e-05,
+      "loss": 3.6681,
+      "step": 16835
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.51438110945734e-05,
+      "loss": 6.1348,
+      "step": 16840
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.510358421497245e-05,
+      "loss": 3.2572,
+      "step": 16845
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.50633573353715e-05,
+      "loss": 4.0573,
+      "step": 16850
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.502313045577055e-05,
+      "loss": 3.3355,
+      "step": 16855
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.498290357616961e-05,
+      "loss": 3.1018,
+      "step": 16860
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.494267669656864e-05,
+      "loss": 2.3866,
+      "step": 16865
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.49024498169677e-05,
+      "loss": 4.0219,
+      "step": 16870
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.486222293736676e-05,
+      "loss": 1.12,
+      "step": 16875
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.48219960577658e-05,
+      "loss": 1.5368,
+      "step": 16880
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.478176917816485e-05,
+      "loss": 3.0138,
+      "step": 16885
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.47415422985639e-05,
+      "loss": 3.6557,
+      "step": 16890
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.470131541896296e-05,
+      "loss": 3.2788,
+      "step": 16895
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.4661088539362e-05,
+      "loss": 3.7785,
+      "step": 16900
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.462086165976105e-05,
+      "loss": 3.4871,
+      "step": 16905
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.458063478016011e-05,
+      "loss": 4.8828,
+      "step": 16910
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.454040790055916e-05,
+      "loss": 3.833,
+      "step": 16915
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.45001810209582e-05,
+      "loss": 1.6004,
+      "step": 16920
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.445995414135726e-05,
+      "loss": 2.4341,
+      "step": 16925
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.441972726175632e-05,
+      "loss": 2.7167,
+      "step": 16930
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.437950038215536e-05,
+      "loss": 3.296,
+      "step": 16935
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.43392735025544e-05,
+      "loss": 4.0896,
+      "step": 16940
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.429904662295346e-05,
+      "loss": 2.8058,
+      "step": 16945
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.425881974335251e-05,
+      "loss": 2.2906,
+      "step": 16950
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.421859286375155e-05,
+      "loss": 3.6217,
+      "step": 16955
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.417836598415061e-05,
+      "loss": 2.6921,
+      "step": 16960
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.413813910454967e-05,
+      "loss": 3.0229,
+      "step": 16965
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.409791222494872e-05,
+      "loss": 2.7896,
+      "step": 16970
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.405768534534776e-05,
+      "loss": 2.8448,
+      "step": 16975
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.401745846574682e-05,
+      "loss": 2.2765,
+      "step": 16980
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.397723158614586e-05,
+      "loss": 3.1928,
+      "step": 16985
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.393700470654492e-05,
+      "loss": 4.5197,
+      "step": 16990
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.389677782694397e-05,
+      "loss": 4.3378,
+      "step": 16995
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.385655094734301e-05,
+      "loss": 3.5922,
+      "step": 17000
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.381632406774207e-05,
+      "loss": 3.5587,
+      "step": 17005
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.377609718814113e-05,
+      "loss": 3.9139,
+      "step": 17010
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.373587030854017e-05,
+      "loss": 2.5823,
+      "step": 17015
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.369564342893922e-05,
+      "loss": 2.5079,
+      "step": 17020
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.365541654933827e-05,
+      "loss": 0.9608,
+      "step": 17025
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.362323504565751e-05,
+      "loss": 2.5257,
+      "step": 17030
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.358300816605657e-05,
+      "loss": 3.4347,
+      "step": 17035
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.354278128645561e-05,
+      "loss": 3.7679,
+      "step": 17040
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.350255440685467e-05,
+      "loss": 3.6029,
+      "step": 17045
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.346232752725371e-05,
+      "loss": 3.6551,
+      "step": 17050
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.342210064765277e-05,
+      "loss": 4.3367,
+      "step": 17055
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.338187376805182e-05,
+      "loss": 3.6209,
+      "step": 17060
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.334164688845086e-05,
+      "loss": 4.4252,
+      "step": 17065
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.330142000884992e-05,
+      "loss": 3.3807,
+      "step": 17070
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.326119312924898e-05,
+      "loss": 3.8533,
+      "step": 17075
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.322096624964801e-05,
+      "loss": 2.6486,
+      "step": 17080
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.318073937004707e-05,
+      "loss": 1.0389,
+      "step": 17085
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.314051249044613e-05,
+      "loss": 4.8477,
+      "step": 17090
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.310028561084517e-05,
+      "loss": 4.098,
+      "step": 17095
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.306005873124421e-05,
+      "loss": 3.5662,
+      "step": 17100
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.301983185164327e-05,
+      "loss": 4.3957,
+      "step": 17105
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.297960497204233e-05,
+      "loss": 1.8946,
+      "step": 17110
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.293937809244136e-05,
+      "loss": 4.3447,
+      "step": 17115
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.289915121284042e-05,
+      "loss": 3.3158,
+      "step": 17120
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.285892433323948e-05,
+      "loss": 1.9162,
+      "step": 17125
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.281869745363852e-05,
+      "loss": 2.8082,
+      "step": 17130
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.277847057403757e-05,
+      "loss": 2.7045,
+      "step": 17135
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.273824369443663e-05,
+      "loss": 4.3832,
+      "step": 17140
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.269801681483568e-05,
+      "loss": 4.2967,
+      "step": 17145
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.265778993523473e-05,
+      "loss": 4.1742,
+      "step": 17150
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.261756305563377e-05,
+      "loss": 4.3498,
+      "step": 17155
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.257733617603283e-05,
+      "loss": 3.0135,
+      "step": 17160
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.253710929643188e-05,
+      "loss": 3.2709,
+      "step": 17165
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.249688241683092e-05,
+      "loss": 2.5106,
+      "step": 17170
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.245665553722998e-05,
+      "loss": 1.6757,
+      "step": 17175
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.241642865762904e-05,
+      "loss": 2.2027,
+      "step": 17180
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.237620177802808e-05,
+      "loss": 3.2238,
+      "step": 17185
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.233597489842713e-05,
+      "loss": 4.7036,
+      "step": 17190
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.229574801882619e-05,
+      "loss": 4.2068,
+      "step": 17195
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.225552113922523e-05,
+      "loss": 3.9623,
+      "step": 17200
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.221529425962429e-05,
+      "loss": 3.6021,
+      "step": 17205
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.217506738002333e-05,
+      "loss": 2.7974,
+      "step": 17210
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.213484050042238e-05,
+      "loss": 3.2512,
+      "step": 17215
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.209461362082144e-05,
+      "loss": 2.4599,
+      "step": 17220
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.20543867412205e-05,
+      "loss": 3.6352,
+      "step": 17225
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.201415986161954e-05,
+      "loss": 1.6141,
+      "step": 17230
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.197393298201858e-05,
+      "loss": 1.3339,
+      "step": 17235
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.193370610241764e-05,
+      "loss": 4.0625,
+      "step": 17240
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.189347922281669e-05,
+      "loss": 3.9266,
+      "step": 17245
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.185325234321573e-05,
+      "loss": 4.1445,
+      "step": 17250
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.181302546361479e-05,
+      "loss": 4.4178,
+      "step": 17255
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.177279858401385e-05,
+      "loss": 5.1043,
+      "step": 17260
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.173257170441289e-05,
+      "loss": 1.8865,
+      "step": 17265
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.169234482481194e-05,
+      "loss": 3.1843,
+      "step": 17270
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.1652117945211e-05,
+      "loss": 3.2057,
+      "step": 17275
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.161189106561005e-05,
+      "loss": 2.154,
+      "step": 17280
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.157166418600909e-05,
+      "loss": 4.2513,
+      "step": 17285
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.153143730640814e-05,
+      "loss": 4.2359,
+      "step": 17290
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.14912104268072e-05,
+      "loss": 3.6398,
+      "step": 17295
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.145098354720625e-05,
+      "loss": 3.352,
+      "step": 17300
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.141075666760529e-05,
+      "loss": 2.9969,
+      "step": 17305
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.137052978800435e-05,
+      "loss": 2.6333,
+      "step": 17310
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.133030290840341e-05,
+      "loss": 3.7852,
+      "step": 17315
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.129007602880244e-05,
+      "loss": 1.6063,
+      "step": 17320
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.12498491492015e-05,
+      "loss": 3.288,
+      "step": 17325
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.120962226960056e-05,
+      "loss": 1.2184,
+      "step": 17330
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.11693953899996e-05,
+      "loss": 2.4909,
+      "step": 17335
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.112916851039864e-05,
+      "loss": 4.035,
+      "step": 17340
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.10889416307977e-05,
+      "loss": 4.6348,
+      "step": 17345
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.104871475119676e-05,
+      "loss": 3.6294,
+      "step": 17350
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.10084878715958e-05,
+      "loss": 4.0395,
+      "step": 17355
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.096826099199485e-05,
+      "loss": 4.607,
+      "step": 17360
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.092803411239391e-05,
+      "loss": 3.9953,
+      "step": 17365
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.088780723279295e-05,
+      "loss": 2.5543,
+      "step": 17370
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.084758035319201e-05,
+      "loss": 2.2387,
+      "step": 17375
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.080735347359106e-05,
+      "loss": 1.8694,
+      "step": 17380
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.07671265939901e-05,
+      "loss": 2.5382,
+      "step": 17385
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.072689971438916e-05,
+      "loss": 5.0727,
+      "step": 17390
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.06866728347882e-05,
+      "loss": 3.9621,
+      "step": 17395
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.064644595518726e-05,
+      "loss": 4.0389,
+      "step": 17400
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.06062190755863e-05,
+      "loss": 3.7287,
+      "step": 17405
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.056599219598536e-05,
+      "loss": 3.1371,
+      "step": 17410
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.052576531638441e-05,
+      "loss": 2.7134,
+      "step": 17415
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.048553843678345e-05,
+      "loss": 3.0964,
+      "step": 17420
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.044531155718251e-05,
+      "loss": 1.5899,
+      "step": 17425
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.040508467758157e-05,
+      "loss": 1.634,
+      "step": 17430
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.036485779798062e-05,
+      "loss": 3.0479,
+      "step": 17435
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.032463091837966e-05,
+      "loss": 4.1072,
+      "step": 17440
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.028440403877872e-05,
+      "loss": 3.6551,
+      "step": 17445
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.024417715917778e-05,
+      "loss": 4.5043,
+      "step": 17450
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.020395027957681e-05,
+      "loss": 3.7246,
+      "step": 17455
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.016372339997587e-05,
+      "loss": 3.2265,
+      "step": 17460
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.012349652037492e-05,
+      "loss": 3.3408,
+      "step": 17465
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.008326964077397e-05,
+      "loss": 3.5267,
+      "step": 17470
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.004304276117301e-05,
+      "loss": 1.8221,
+      "step": 17475
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.000281588157207e-05,
+      "loss": 1.5264,
+      "step": 17480
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.996258900197113e-05,
+      "loss": 2.7208,
+      "step": 17485
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.992236212237016e-05,
+      "loss": 3.7732,
+      "step": 17490
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.988213524276922e-05,
+      "loss": 4.3205,
+      "step": 17495
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.984190836316828e-05,
+      "loss": 4.0115,
+      "step": 17500
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.980168148356732e-05,
+      "loss": 3.7114,
+      "step": 17505
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.976145460396637e-05,
+      "loss": 2.9249,
+      "step": 17510
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.972122772436543e-05,
+      "loss": 1.8061,
+      "step": 17515
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.968100084476447e-05,
+      "loss": 1.9254,
+      "step": 17520
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.964077396516353e-05,
+      "loss": 0.8883,
+      "step": 17525
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.960054708556257e-05,
+      "loss": 0.7865,
+      "step": 17530
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.956032020596163e-05,
+      "loss": 1.5326,
+      "step": 17535
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.952009332636068e-05,
+      "loss": 3.7009,
+      "step": 17540
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.947986644675972e-05,
+      "loss": 3.5918,
+      "step": 17545
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.943963956715878e-05,
+      "loss": 3.55,
+      "step": 17550
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.939941268755782e-05,
+      "loss": 2.8442,
+      "step": 17555
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.935918580795688e-05,
+      "loss": 3.8076,
+      "step": 17560
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.931895892835593e-05,
+      "loss": 2.268,
+      "step": 17565
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.927873204875498e-05,
+      "loss": 1.8671,
+      "step": 17570
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.923850516915403e-05,
+      "loss": 3.1274,
+      "step": 17575
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.919827828955309e-05,
+      "loss": 1.8582,
+      "step": 17580
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.915805140995213e-05,
+      "loss": 1.7748,
+      "step": 17585
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.911782453035118e-05,
+      "loss": 3.6909,
+      "step": 17590
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.907759765075024e-05,
+      "loss": 4.9395,
+      "step": 17595
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.90373707711493e-05,
+      "loss": 5.6717,
+      "step": 17600
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.899714389154834e-05,
+      "loss": 3.5836,
+      "step": 17605
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.895691701194738e-05,
+      "loss": 4.4979,
+      "step": 17610
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.891669013234644e-05,
+      "loss": 2.4373,
+      "step": 17615
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.887646325274549e-05,
+      "loss": 2.756,
+      "step": 17620
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.883623637314453e-05,
+      "loss": 1.4593,
+      "step": 17625
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.879600949354359e-05,
+      "loss": 1.1477,
+      "step": 17630
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.875578261394265e-05,
+      "loss": 1.9229,
+      "step": 17635
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.871555573434169e-05,
+      "loss": 4.017,
+      "step": 17640
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.867532885474074e-05,
+      "loss": 5.0738,
+      "step": 17645
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.86351019751398e-05,
+      "loss": 3.6027,
+      "step": 17650
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.859487509553885e-05,
+      "loss": 3.1089,
+      "step": 17655
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.855464821593788e-05,
+      "loss": 3.3415,
+      "step": 17660
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.851442133633694e-05,
+      "loss": 3.0649,
+      "step": 17665
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.8474194456736e-05,
+      "loss": 1.925,
+      "step": 17670
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.843396757713505e-05,
+      "loss": 2.909,
+      "step": 17675
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.839374069753409e-05,
+      "loss": 3.0972,
+      "step": 17680
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.835351381793315e-05,
+      "loss": 1.8968,
+      "step": 17685
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.831328693833219e-05,
+      "loss": 4.1662,
+      "step": 17690
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.827306005873125e-05,
+      "loss": 3.5768,
+      "step": 17695
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.82328331791303e-05,
+      "loss": 3.7977,
+      "step": 17700
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.819260629952935e-05,
+      "loss": 2.5532,
+      "step": 17705
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.81523794199284e-05,
+      "loss": 2.4193,
+      "step": 17710
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.811215254032744e-05,
+      "loss": 2.7911,
+      "step": 17715
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.80719256607265e-05,
+      "loss": 1.1723,
+      "step": 17720
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.803169878112555e-05,
+      "loss": 2.1002,
+      "step": 17725
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.79914719015246e-05,
+      "loss": 1.357,
+      "step": 17730
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.795124502192365e-05,
+      "loss": 2.7562,
+      "step": 17735
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.791101814232271e-05,
+      "loss": 3.9273,
+      "step": 17740
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.787079126272175e-05,
+      "loss": 4.8758,
+      "step": 17745
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.783056438312081e-05,
+      "loss": 3.9227,
+      "step": 17750
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.779033750351986e-05,
+      "loss": 3.1739,
+      "step": 17755
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.77501106239189e-05,
+      "loss": 3.0794,
+      "step": 17760
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.770988374431796e-05,
+      "loss": 2.0279,
+      "step": 17765
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.766965686471702e-05,
+      "loss": 2.2346,
+      "step": 17770
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.762942998511606e-05,
+      "loss": 2.6593,
+      "step": 17775
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.75892031055151e-05,
+      "loss": 0.953,
+      "step": 17780
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.754897622591416e-05,
+      "loss": 2.3692,
+      "step": 17785
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.750874934631321e-05,
+      "loss": 4.9299,
+      "step": 17790
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.746852246671225e-05,
+      "loss": 4.4605,
+      "step": 17795
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.742829558711131e-05,
+      "loss": 4.1517,
+      "step": 17800
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.738806870751037e-05,
+      "loss": 2.3445,
+      "step": 17805
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.734784182790941e-05,
+      "loss": 4.102,
+      "step": 17810
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.730761494830846e-05,
+      "loss": 2.3169,
+      "step": 17815
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.726738806870752e-05,
+      "loss": 3.781,
+      "step": 17820
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.722716118910656e-05,
+      "loss": 1.5209,
+      "step": 17825
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.718693430950561e-05,
+      "loss": 2.827,
+      "step": 17830
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.714670742990467e-05,
+      "loss": 1.0628,
+      "step": 17835
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.710648055030372e-05,
+      "loss": 4.5979,
+      "step": 17840
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.706625367070277e-05,
+      "loss": 3.8928,
+      "step": 17845
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.702602679110181e-05,
+      "loss": 4.7121,
+      "step": 17850
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.698579991150087e-05,
+      "loss": 3.5897,
+      "step": 17855
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.694557303189992e-05,
+      "loss": 2.4365,
+      "step": 17860
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.690534615229896e-05,
+      "loss": 3.8054,
+      "step": 17865
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.686511927269802e-05,
+      "loss": 3.1852,
+      "step": 17870
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.682489239309708e-05,
+      "loss": 1.203,
+      "step": 17875
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.678466551349612e-05,
+      "loss": 0.7864,
+      "step": 17880
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.674443863389517e-05,
+      "loss": 1.6607,
+      "step": 17885
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.670421175429422e-05,
+      "loss": 4.7062,
+      "step": 17890
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.666398487469327e-05,
+      "loss": 4.7527,
+      "step": 17895
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.662375799509233e-05,
+      "loss": 3.8579,
+      "step": 17900
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.658353111549137e-05,
+      "loss": 3.7981,
+      "step": 17905
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.654330423589043e-05,
+      "loss": 2.9869,
+      "step": 17910
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.650307735628948e-05,
+      "loss": 1.7792,
+      "step": 17915
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.646285047668853e-05,
+      "loss": 4.4921,
+      "step": 17920
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.642262359708758e-05,
+      "loss": 3.5635,
+      "step": 17925
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.638239671748662e-05,
+      "loss": 2.6001,
+      "step": 17930
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.634216983788568e-05,
+      "loss": 1.0408,
+      "step": 17935
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.630194295828473e-05,
+      "loss": 4.8291,
+      "step": 17940
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.626171607868378e-05,
+      "loss": 4.6105,
+      "step": 17945
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.622148919908283e-05,
+      "loss": 3.6875,
+      "step": 17950
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.618126231948189e-05,
+      "loss": 3.7762,
+      "step": 17955
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.614103543988093e-05,
+      "loss": 3.6926,
+      "step": 17960
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.610080856027998e-05,
+      "loss": 4.0579,
+      "step": 17965
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.606058168067903e-05,
+      "loss": 2.8567,
+      "step": 17970
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.602035480107809e-05,
+      "loss": 2.1443,
+      "step": 17975
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.598012792147714e-05,
+      "loss": 3.152,
+      "step": 17980
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.593990104187618e-05,
+      "loss": 1.3468,
+      "step": 17985
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.589967416227524e-05,
+      "loss": 4.0677,
+      "step": 17990
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.585944728267429e-05,
+      "loss": 4.3828,
+      "step": 17995
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.581922040307333e-05,
+      "loss": 4.085,
+      "step": 18000
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.577899352347239e-05,
+      "loss": 3.9857,
+      "step": 18005
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.573876664387145e-05,
+      "loss": 4.3557,
+      "step": 18010
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.569853976427048e-05,
+      "loss": 3.3637,
+      "step": 18015
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.565831288466954e-05,
+      "loss": 3.1275,
+      "step": 18020
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.56180860050686e-05,
+      "loss": 1.9914,
+      "step": 18025
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.557785912546764e-05,
+      "loss": 1.9206,
+      "step": 18030
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.553763224586668e-05,
+      "loss": 2.7614,
+      "step": 18035
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.549740536626574e-05,
+      "loss": 3.8361,
+      "step": 18040
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.54571784866648e-05,
+      "loss": 4.5418,
+      "step": 18045
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.541695160706384e-05,
+      "loss": 3.9312,
+      "step": 18050
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.537672472746289e-05,
+      "loss": 2.5614,
+      "step": 18055
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.533649784786195e-05,
+      "loss": 2.4895,
+      "step": 18060
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.529627096826099e-05,
+      "loss": 1.8234,
+      "step": 18065
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.525604408866005e-05,
+      "loss": 2.198,
+      "step": 18070
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.52158172090591e-05,
+      "loss": 1.9346,
+      "step": 18075
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.517559032945815e-05,
+      "loss": 1.823,
+      "step": 18080
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.51353634498572e-05,
+      "loss": 1.2123,
+      "step": 18085
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.509513657025624e-05,
+      "loss": 3.9965,
+      "step": 18090
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.50549096906553e-05,
+      "loss": 4.3736,
+      "step": 18095
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.501468281105435e-05,
+      "loss": 3.9881,
+      "step": 18100
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.49744559314534e-05,
+      "loss": 3.8488,
+      "step": 18105
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.493422905185245e-05,
+      "loss": 4.2275,
+      "step": 18110
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.48940021722515e-05,
+      "loss": 4.3555,
+      "step": 18115
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.485377529265055e-05,
+      "loss": 3.4964,
+      "step": 18120
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.481354841304961e-05,
+      "loss": 2.727,
+      "step": 18125
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.477332153344865e-05,
+      "loss": 3.9642,
+      "step": 18130
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.47330946538477e-05,
+      "loss": 1.054,
+      "step": 18135
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.469286777424676e-05,
+      "loss": 4.3533,
+      "step": 18140
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.465264089464582e-05,
+      "loss": 3.6725,
+      "step": 18145
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.461241401504485e-05,
+      "loss": 3.9557,
+      "step": 18150
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.45721871354439e-05,
+      "loss": 4.5719,
+      "step": 18155
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.453196025584296e-05,
+      "loss": 3.3895,
+      "step": 18160
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.449173337624201e-05,
+      "loss": 2.3612,
+      "step": 18165
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.445150649664105e-05,
+      "loss": 3.5728,
+      "step": 18170
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.441127961704011e-05,
+      "loss": 1.809,
+      "step": 18175
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.437105273743917e-05,
+      "loss": 2.2907,
+      "step": 18180
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.43308258578382e-05,
+      "loss": 3.9573,
+      "step": 18185
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.429059897823726e-05,
+      "loss": 3.91,
+      "step": 18190
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.425037209863632e-05,
+      "loss": 3.9791,
+      "step": 18195
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.421014521903536e-05,
+      "loss": 4.3547,
+      "step": 18200
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.41699183394344e-05,
+      "loss": 4.3502,
+      "step": 18205
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.412969145983346e-05,
+      "loss": 3.3553,
+      "step": 18210
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.408946458023252e-05,
+      "loss": 2.7739,
+      "step": 18215
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.404923770063157e-05,
+      "loss": 2.2082,
+      "step": 18220
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.400901082103061e-05,
+      "loss": 1.1458,
+      "step": 18225
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.396878394142967e-05,
+      "loss": 2.6004,
+      "step": 18230
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.392855706182871e-05,
+      "loss": 2.9328,
+      "step": 18235
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.388833018222776e-05,
+      "loss": 4.5781,
+      "step": 18240
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.384810330262682e-05,
+      "loss": 4.5582,
+      "step": 18245
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.380787642302588e-05,
+      "loss": 3.7113,
+      "step": 18250
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.376764954342492e-05,
+      "loss": 3.8949,
+      "step": 18255
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.372742266382397e-05,
+      "loss": 4.1025,
+      "step": 18260
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.368719578422302e-05,
+      "loss": 4.8066,
+      "step": 18265
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.364696890462207e-05,
+      "loss": 3.1322,
+      "step": 18270
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.360674202502113e-05,
+      "loss": 2.281,
+      "step": 18275
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.356651514542017e-05,
+      "loss": 1.8965,
+      "step": 18280
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.352628826581922e-05,
+      "loss": 2.0165,
+      "step": 18285
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.348606138621827e-05,
+      "loss": 4.3139,
+      "step": 18290
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.344583450661733e-05,
+      "loss": 3.5159,
+      "step": 18295
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.340560762701638e-05,
+      "loss": 3.6641,
+      "step": 18300
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.336538074741542e-05,
+      "loss": 4.3941,
+      "step": 18305
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.332515386781448e-05,
+      "loss": 4.0872,
+      "step": 18310
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.328492698821352e-05,
+      "loss": 2.6069,
+      "step": 18315
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.324470010861257e-05,
+      "loss": 2.309,
+      "step": 18320
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.320447322901163e-05,
+      "loss": 2.1683,
+      "step": 18325
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.316424634941069e-05,
+      "loss": 1.5804,
+      "step": 18330
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.312401946980973e-05,
+      "loss": 2.2497,
+      "step": 18335
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.308379259020878e-05,
+      "loss": 3.6252,
+      "step": 18340
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.304356571060783e-05,
+      "loss": 4.0513,
+      "step": 18345
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.300333883100689e-05,
+      "loss": 3.3787,
+      "step": 18350
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.296311195140592e-05,
+      "loss": 3.8106,
+      "step": 18355
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.292288507180498e-05,
+      "loss": 4.0287,
+      "step": 18360
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.288265819220404e-05,
+      "loss": 2.5771,
+      "step": 18365
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.284243131260308e-05,
+      "loss": 3.1564,
+      "step": 18370
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.280220443300213e-05,
+      "loss": 0.9893,
+      "step": 18375
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.276197755340119e-05,
+      "loss": 1.7055,
+      "step": 18380
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.272175067380025e-05,
+      "loss": 2.0058,
+      "step": 18385
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.268152379419928e-05,
+      "loss": 4.6082,
+      "step": 18390
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.264129691459833e-05,
+      "loss": 4.1299,
+      "step": 18395
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.260107003499739e-05,
+      "loss": 2.9067,
+      "step": 18400
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.256084315539644e-05,
+      "loss": 4.0816,
+      "step": 18405
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.252061627579548e-05,
+      "loss": 4.8357,
+      "step": 18410
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.248038939619454e-05,
+      "loss": 3.3007,
+      "step": 18415
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.24401625165936e-05,
+      "loss": 2.1316,
+      "step": 18420
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.239993563699264e-05,
+      "loss": 3.2057,
+      "step": 18425
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.235970875739169e-05,
+      "loss": 1.6077,
+      "step": 18430
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.231948187779075e-05,
+      "loss": 1.3006,
+      "step": 18435
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.227925499818979e-05,
+      "loss": 4.4477,
+      "step": 18440
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.223902811858885e-05,
+      "loss": 3.8949,
+      "step": 18445
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.21988012389879e-05,
+      "loss": 4.133,
+      "step": 18450
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.215857435938694e-05,
+      "loss": 4.193,
+      "step": 18455
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.2118347479786e-05,
+      "loss": 4.1554,
+      "step": 18460
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.207812060018504e-05,
+      "loss": 3.3928,
+      "step": 18465
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.20378937205841e-05,
+      "loss": 2.6005,
+      "step": 18470
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.199766684098314e-05,
+      "loss": 1.3332,
+      "step": 18475
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.19574399613822e-05,
+      "loss": 1.9054,
+      "step": 18480
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.191721308178125e-05,
+      "loss": 1.7774,
+      "step": 18485
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.187698620218029e-05,
+      "loss": 4.46,
+      "step": 18490
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.183675932257935e-05,
+      "loss": 4.0219,
+      "step": 18495
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.179653244297841e-05,
+      "loss": 2.6053,
+      "step": 18500
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.175630556337745e-05,
+      "loss": 3.7852,
+      "step": 18505
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.17160786837765e-05,
+      "loss": 3.012,
+      "step": 18510
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.167585180417556e-05,
+      "loss": 4.4118,
+      "step": 18515
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.163562492457461e-05,
+      "loss": 3.0597,
+      "step": 18520
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.159539804497365e-05,
+      "loss": 2.179,
+      "step": 18525
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.15551711653727e-05,
+      "loss": 1.4966,
+      "step": 18530
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.151494428577176e-05,
+      "loss": 1.3322,
+      "step": 18535
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.147471740617081e-05,
+      "loss": 3.75,
+      "step": 18540
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.143449052656985e-05,
+      "loss": 4.7051,
+      "step": 18545
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.139426364696891e-05,
+      "loss": 5.2428,
+      "step": 18550
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.135403676736797e-05,
+      "loss": 3.5576,
+      "step": 18555
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.1313809887767e-05,
+      "loss": 3.1026,
+      "step": 18560
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.127358300816606e-05,
+      "loss": 4.2556,
+      "step": 18565
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.123335612856512e-05,
+      "loss": 1.9716,
+      "step": 18570
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.119312924896416e-05,
+      "loss": 1.7601,
+      "step": 18575
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.11529023693632e-05,
+      "loss": 3.0333,
+      "step": 18580
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.111267548976226e-05,
+      "loss": 1.5847,
+      "step": 18585
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.107244861016131e-05,
+      "loss": 3.682,
+      "step": 18590
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.1032221730560366e-05,
+      "loss": 4.1221,
+      "step": 18595
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.099199485095941e-05,
+      "loss": 4.0278,
+      "step": 18600
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.095176797135846e-05,
+      "loss": 2.6338,
+      "step": 18605
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.091154109175752e-05,
+      "loss": 3.3207,
+      "step": 18610
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.087131421215656e-05,
+      "loss": 3.9458,
+      "step": 18615
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.083108733255562e-05,
+      "loss": 2.4551,
+      "step": 18620
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.079086045295467e-05,
+      "loss": 1.0928,
+      "step": 18625
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.075063357335372e-05,
+      "loss": 1.2878,
+      "step": 18630
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.0710406693752764e-05,
+      "loss": 3.1286,
+      "step": 18635
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.0670179814151816e-05,
+      "loss": 4.3492,
+      "step": 18640
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.0629952934550874e-05,
+      "loss": 4.4941,
+      "step": 18645
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.0589726054949926e-05,
+      "loss": 3.3197,
+      "step": 18650
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.054949917534897e-05,
+      "loss": 2.8205,
+      "step": 18655
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.050927229574802e-05,
+      "loss": 2.2681,
+      "step": 18660
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.046904541614707e-05,
+      "loss": 2.6547,
+      "step": 18665
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.0428818536546125e-05,
+      "loss": 3.5173,
+      "step": 18670
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.038859165694517e-05,
+      "loss": 1.8866,
+      "step": 18675
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.034836477734422e-05,
+      "loss": 2.6423,
+      "step": 18680
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.030813789774328e-05,
+      "loss": 1.6941,
+      "step": 18685
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.026791101814232e-05,
+      "loss": 3.9561,
+      "step": 18690
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.0227684138541375e-05,
+      "loss": 4.3443,
+      "step": 18695
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.018745725894043e-05,
+      "loss": 3.6342,
+      "step": 18700
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.014723037933948e-05,
+      "loss": 3.3798,
+      "step": 18705
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.010700349973852e-05,
+      "loss": 2.8606,
+      "step": 18710
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.0066776620137574e-05,
+      "loss": 2.683,
+      "step": 18715
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.002654974053663e-05,
+      "loss": 1.8671,
+      "step": 18720
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.9986322860935684e-05,
+      "loss": 4.4889,
+      "step": 18725
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.994609598133473e-05,
+      "loss": 2.7197,
+      "step": 18730
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.990586910173378e-05,
+      "loss": 3.2302,
+      "step": 18735
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.986564222213283e-05,
+      "loss": 3.9961,
+      "step": 18740
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.982541534253189e-05,
+      "loss": 3.467,
+      "step": 18745
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.978518846293093e-05,
+      "loss": 4.3445,
+      "step": 18750
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.9744961583329986e-05,
+      "loss": 4.9615,
+      "step": 18755
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.970473470372904e-05,
+      "loss": 3.1598,
+      "step": 18760
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.966450782412808e-05,
+      "loss": 3.2324,
+      "step": 18765
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.9624280944527134e-05,
+      "loss": 3.9107,
+      "step": 18770
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.9584054064926185e-05,
+      "loss": 1.1585,
+      "step": 18775
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.9543827185325243e-05,
+      "loss": 0.6935,
+      "step": 18780
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.950360030572428e-05,
+      "loss": 1.9886,
+      "step": 18785
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.946337342612334e-05,
+      "loss": 6.2145,
+      "step": 18790
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.942314654652239e-05,
+      "loss": 4.3086,
+      "step": 18795
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.938291966692144e-05,
+      "loss": 3.2976,
+      "step": 18800
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.934269278732049e-05,
+      "loss": 3.9161,
+      "step": 18805
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.930246590771954e-05,
+      "loss": 3.1264,
+      "step": 18810
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.92622390281186e-05,
+      "loss": 4.4698,
+      "step": 18815
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.922201214851765e-05,
+      "loss": 3.101,
+      "step": 18820
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.9181785268916686e-05,
+      "loss": 1.7798,
+      "step": 18825
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.9141558389315745e-05,
+      "loss": 2.7562,
+      "step": 18830
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.9101331509714796e-05,
+      "loss": 2.8743,
+      "step": 18835
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.906110463011385e-05,
+      "loss": 4.3965,
+      "step": 18840
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.902087775051289e-05,
+      "loss": 3.4023,
+      "step": 18845
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.8980650870911944e-05,
+      "loss": 4.3494,
+      "step": 18850
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.8940423991311e-05,
+      "loss": 2.9739,
+      "step": 18855
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.890019711171004e-05,
+      "loss": 4.0516,
+      "step": 18860
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.88599702321091e-05,
+      "loss": 2.3647,
+      "step": 18865
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.881974335250815e-05,
+      "loss": 2.4891,
+      "step": 18870
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.87795164729072e-05,
+      "loss": 2.7286,
+      "step": 18875
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.8739289593306246e-05,
+      "loss": 2.3304,
+      "step": 18880
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.86990627137053e-05,
+      "loss": 1.6823,
+      "step": 18885
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.8658835834104355e-05,
+      "loss": 4.1017,
+      "step": 18890
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.861860895450341e-05,
+      "loss": 4.4844,
+      "step": 18895
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.857838207490245e-05,
+      "loss": 3.8736,
+      "step": 18900
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.85381551953015e-05,
+      "loss": 3.4389,
+      "step": 18905
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.8497928315700554e-05,
+      "loss": 4.6137,
+      "step": 18910
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.845770143609961e-05,
+      "loss": 3.7057,
+      "step": 18915
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.841747455649865e-05,
+      "loss": 1.9062,
+      "step": 18920
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.837724767689771e-05,
+      "loss": 2.6992,
+      "step": 18925
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.833702079729676e-05,
+      "loss": 2.4897,
+      "step": 18930
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.8296793917695805e-05,
+      "loss": 2.3581,
+      "step": 18935
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.8256567038094856e-05,
+      "loss": 4.217,
+      "step": 18940
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.821634015849391e-05,
+      "loss": 3.4085,
+      "step": 18945
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.8176113278892966e-05,
+      "loss": 3.5184,
+      "step": 18950
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.8135886399292004e-05,
+      "loss": 3.6127,
+      "step": 18955
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.809565951969106e-05,
+      "loss": 3.4811,
+      "step": 18960
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.8055432640090114e-05,
+      "loss": 3.3875,
+      "step": 18965
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.8015205760489165e-05,
+      "loss": 2.1214,
+      "step": 18970
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.797497888088821e-05,
+      "loss": 3.5039,
+      "step": 18975
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.793475200128726e-05,
+      "loss": 1.3013,
+      "step": 18980
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.789452512168631e-05,
+      "loss": 1.1169,
+      "step": 18985
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.785429824208537e-05,
+      "loss": 3.883,
+      "step": 18990
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.781407136248441e-05,
+      "loss": 4.4398,
+      "step": 18995
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.777384448288347e-05,
+      "loss": 2.2548,
+      "step": 19000
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.773361760328252e-05,
+      "loss": 4.5357,
+      "step": 19005
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.7693390723681563e-05,
+      "loss": 3.7588,
+      "step": 19010
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.7653163844080615e-05,
+      "loss": 1.6498,
+      "step": 19015
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.7612936964479666e-05,
+      "loss": 3.7896,
+      "step": 19020
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.7572710084878725e-05,
+      "loss": 1.6262,
+      "step": 19025
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.753248320527776e-05,
+      "loss": 0.6316,
+      "step": 19030
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.749225632567682e-05,
+      "loss": 0.6665,
+      "step": 19035
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.745202944607587e-05,
+      "loss": 4.3615,
+      "step": 19040
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.7411802566474924e-05,
+      "loss": 4.8541,
+      "step": 19045
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.737157568687397e-05,
+      "loss": 3.6393,
+      "step": 19050
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.733134880727302e-05,
+      "loss": 4.5691,
+      "step": 19055
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.729112192767208e-05,
+      "loss": 3.1455,
+      "step": 19060
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.725089504807113e-05,
+      "loss": 3.8037,
+      "step": 19065
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.7210668168470174e-05,
+      "loss": 2.758,
+      "step": 19070
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.7170441288869226e-05,
+      "loss": 2.8246,
+      "step": 19075
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.713021440926828e-05,
+      "loss": 2.058,
+      "step": 19080
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.708998752966732e-05,
+      "loss": 1.0103,
+      "step": 19085
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.704976065006637e-05,
+      "loss": 3.3998,
+      "step": 19090
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.700953377046543e-05,
+      "loss": 3.6197,
+      "step": 19095
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.696930689086448e-05,
+      "loss": 3.9162,
+      "step": 19100
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.692908001126353e-05,
+      "loss": 4.3734,
+      "step": 19105
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.688885313166258e-05,
+      "loss": 3.8322,
+      "step": 19110
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.684862625206163e-05,
+      "loss": 2.6953,
+      "step": 19115
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.680839937246068e-05,
+      "loss": 2.9081,
+      "step": 19120
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.676817249285973e-05,
+      "loss": 2.9252,
+      "step": 19125
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.672794561325878e-05,
+      "loss": 2.051,
+      "step": 19130
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.6687718733657837e-05,
+      "loss": 4.5108,
+      "step": 19135
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.664749185405689e-05,
+      "loss": 4.5414,
+      "step": 19140
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.660726497445593e-05,
+      "loss": 4.4689,
+      "step": 19145
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.6567038094854984e-05,
+      "loss": 3.826,
+      "step": 19150
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.6526811215254036e-05,
+      "loss": 2.8979,
+      "step": 19155
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.648658433565308e-05,
+      "loss": 3.5128,
+      "step": 19160
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.644635745605213e-05,
+      "loss": 2.2047,
+      "step": 19165
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.640613057645119e-05,
+      "loss": 2.9307,
+      "step": 19170
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.636590369685024e-05,
+      "loss": 0.9308,
+      "step": 19175
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.6325676817249286e-05,
+      "loss": 1.3805,
+      "step": 19180
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.628544993764834e-05,
+      "loss": 3.8648,
+      "step": 19185
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.624522305804739e-05,
+      "loss": 4.8469,
+      "step": 19190
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.620499617844645e-05,
+      "loss": 4.4084,
+      "step": 19195
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.6164769298845485e-05,
+      "loss": 4.0066,
+      "step": 19200
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.6124542419244544e-05,
+      "loss": 2.9517,
+      "step": 19205
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.6084315539643595e-05,
+      "loss": 3.6738,
+      "step": 19210
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.6044088660042647e-05,
+      "loss": 2.7926,
+      "step": 19215
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.600386178044169e-05,
+      "loss": 1.8658,
+      "step": 19220
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.596363490084074e-05,
+      "loss": 2.3674,
+      "step": 19225
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.59234080212398e-05,
+      "loss": 3.3319,
+      "step": 19230
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.588318114163884e-05,
+      "loss": 2.3487,
+      "step": 19235
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.58429542620379e-05,
+      "loss": 4.4271,
+      "step": 19240
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.580272738243695e-05,
+      "loss": 5.4869,
+      "step": 19245
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.5762500502836e-05,
+      "loss": 3.9018,
+      "step": 19250
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.5722273623235045e-05,
+      "loss": 4.5006,
+      "step": 19255
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.5682046743634096e-05,
+      "loss": 4.2176,
+      "step": 19260
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.5641819864033154e-05,
+      "loss": 2.4257,
+      "step": 19265
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.5601592984432206e-05,
+      "loss": 2.4731,
+      "step": 19270
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.556136610483125e-05,
+      "loss": 3.5372,
+      "step": 19275
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.55211392252303e-05,
+      "loss": 1.971,
+      "step": 19280
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.5480912345629353e-05,
+      "loss": 2.1551,
+      "step": 19285
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.5440685466028405e-05,
+      "loss": 3.9453,
+      "step": 19290
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.540045858642745e-05,
+      "loss": 4.4363,
+      "step": 19295
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.53602317068265e-05,
+      "loss": 3.8611,
+      "step": 19300
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.532000482722556e-05,
+      "loss": 3.0589,
+      "step": 19305
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.52797779476246e-05,
+      "loss": 3.3494,
+      "step": 19310
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.5239551068023656e-05,
+      "loss": 2.2144,
+      "step": 19315
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.519932418842271e-05,
+      "loss": 4.4766,
+      "step": 19320
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.515909730882176e-05,
+      "loss": 3.267,
+      "step": 19325
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.51188704292208e-05,
+      "loss": 2.9234,
+      "step": 19330
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.5078643549619855e-05,
+      "loss": 3.8816,
+      "step": 19335
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.503841667001891e-05,
+      "loss": 4.1756,
+      "step": 19340
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.4998189790417964e-05,
+      "loss": 4.3277,
+      "step": 19345
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.495796291081701e-05,
+      "loss": 4.1113,
+      "step": 19350
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.491773603121606e-05,
+      "loss": 3.8789,
+      "step": 19355
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.487750915161511e-05,
+      "loss": 2.3336,
+      "step": 19360
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.483728227201417e-05,
+      "loss": 3.0759,
+      "step": 19365
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.479705539241321e-05,
+      "loss": 2.9986,
+      "step": 19370
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.4756828512812266e-05,
+      "loss": 2.5309,
+      "step": 19375
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.471660163321132e-05,
+      "loss": 2.5339,
+      "step": 19380
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.467637475361036e-05,
+      "loss": 2.0254,
+      "step": 19385
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.4636147874009414e-05,
+      "loss": 4.4572,
+      "step": 19390
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.4595920994408465e-05,
+      "loss": 4.1877,
+      "step": 19395
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.4555694114807524e-05,
+      "loss": 3.7068,
+      "step": 19400
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.451546723520656e-05,
+      "loss": 3.5297,
+      "step": 19405
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.447524035560562e-05,
+      "loss": 3.059,
+      "step": 19410
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.443501347600467e-05,
+      "loss": 3.3376,
+      "step": 19415
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.439478659640372e-05,
+      "loss": 2.6208,
+      "step": 19420
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.435455971680277e-05,
+      "loss": 0.5573,
+      "step": 19425
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.431433283720182e-05,
+      "loss": 0.388,
+      "step": 19430
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.427410595760087e-05,
+      "loss": 2.8777,
+      "step": 19435
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.423387907799993e-05,
+      "loss": 4.8811,
+      "step": 19440
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.4193652198398967e-05,
+      "loss": 3.5342,
+      "step": 19445
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.4153425318798025e-05,
+      "loss": 3.8281,
+      "step": 19450
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.4113198439197076e-05,
+      "loss": 4.1465,
+      "step": 19455
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.407297155959612e-05,
+      "loss": 3.8576,
+      "step": 19460
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.403274467999517e-05,
+      "loss": 3.6264,
+      "step": 19465
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.3992517800394224e-05,
+      "loss": 3.2877,
+      "step": 19470
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.395229092079328e-05,
+      "loss": 2.1062,
+      "step": 19475
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.391206404119232e-05,
+      "loss": 1.655,
+      "step": 19480
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.387183716159138e-05,
+      "loss": 1.7521,
+      "step": 19485
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.383161028199043e-05,
+      "loss": 4.1643,
+      "step": 19490
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.379138340238948e-05,
+      "loss": 4.5741,
+      "step": 19495
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.3751156522788526e-05,
+      "loss": 4.5176,
+      "step": 19500
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.371092964318758e-05,
+      "loss": 2.9813,
+      "step": 19505
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.3670702763586636e-05,
+      "loss": 3.9217,
+      "step": 19510
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.363047588398569e-05,
+      "loss": 2.6086,
+      "step": 19515
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.359024900438473e-05,
+      "loss": 3.6834,
+      "step": 19520
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.355002212478378e-05,
+      "loss": 3.0846,
+      "step": 19525
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.3509795245182835e-05,
+      "loss": 2.2924,
+      "step": 19530
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.346956836558188e-05,
+      "loss": 3.0072,
+      "step": 19535
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.342934148598093e-05,
+      "loss": 3.6783,
+      "step": 19540
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.338911460637999e-05,
+      "loss": 3.4266,
+      "step": 19545
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.334888772677904e-05,
+      "loss": 4.5687,
+      "step": 19550
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.3308660847178085e-05,
+      "loss": 3.6299,
+      "step": 19555
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.326843396757714e-05,
+      "loss": 4.2918,
+      "step": 19560
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.322820708797619e-05,
+      "loss": 3.1556,
+      "step": 19565
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.3187980208375246e-05,
+      "loss": 1.7684,
+      "step": 19570
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.3147753328774284e-05,
+      "loss": 1.7001,
+      "step": 19575
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.310752644917334e-05,
+      "loss": 2.9054,
+      "step": 19580
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.3067299569572394e-05,
+      "loss": 3.3056,
+      "step": 19585
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.3027072689971446e-05,
+      "loss": 4.6641,
+      "step": 19590
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.298684581037049e-05,
+      "loss": 4.5357,
+      "step": 19595
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.294661893076954e-05,
+      "loss": 4.2678,
+      "step": 19600
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.290639205116859e-05,
+      "loss": 3.8678,
+      "step": 19605
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.286616517156764e-05,
+      "loss": 2.2252,
+      "step": 19610
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.282593829196669e-05,
+      "loss": 2.7532,
+      "step": 19615
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.278571141236575e-05,
+      "loss": 3.1894,
+      "step": 19620
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.27454845327648e-05,
+      "loss": 1.7801,
+      "step": 19625
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.2705257653163844e-05,
+      "loss": 0.9882,
+      "step": 19630
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.2665030773562895e-05,
+      "loss": 1.3804,
+      "step": 19635
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.262480389396195e-05,
+      "loss": 4.2154,
+      "step": 19640
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.2584577014361005e-05,
+      "loss": 4.8629,
+      "step": 19645
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.254435013476004e-05,
+      "loss": 2.9258,
+      "step": 19650
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.25041232551591e-05,
+      "loss": 2.9222,
+      "step": 19655
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.246389637555815e-05,
+      "loss": 3.0917,
+      "step": 19660
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.2423669495957204e-05,
+      "loss": 2.4604,
+      "step": 19665
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.238344261635625e-05,
+      "loss": 1.6897,
+      "step": 19670
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.23432157367553e-05,
+      "loss": 1.0687,
+      "step": 19675
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.230298885715436e-05,
+      "loss": 1.2178,
+      "step": 19680
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.2262761977553396e-05,
+      "loss": 2.0848,
+      "step": 19685
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.2222535097952455e-05,
+      "loss": 4.2998,
+      "step": 19690
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.2182308218351506e-05,
+      "loss": 5.0777,
+      "step": 19695
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.214208133875056e-05,
+      "loss": 3.448,
+      "step": 19700
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.21018544591496e-05,
+      "loss": 3.3208,
+      "step": 19705
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.2061627579548654e-05,
+      "loss": 2.9604,
+      "step": 19710
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.202140069994771e-05,
+      "loss": 2.4406,
+      "step": 19715
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.198117382034676e-05,
+      "loss": 1.6549,
+      "step": 19720
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.194094694074581e-05,
+      "loss": 1.4544,
+      "step": 19725
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.190072006114486e-05,
+      "loss": 2.5774,
+      "step": 19730
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.186049318154391e-05,
+      "loss": 2.1335,
+      "step": 19735
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.182026630194296e-05,
+      "loss": 4.3416,
+      "step": 19740
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.178003942234201e-05,
+      "loss": 3.4145,
+      "step": 19745
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.173981254274106e-05,
+      "loss": 3.9055,
+      "step": 19750
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.169958566314012e-05,
+      "loss": 3.7634,
+      "step": 19755
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.1659358783539155e-05,
+      "loss": 3.4955,
+      "step": 19760
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.161913190393821e-05,
+      "loss": 1.8993,
+      "step": 19765
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.1578905024337264e-05,
+      "loss": 3.337,
+      "step": 19770
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.1538678144736316e-05,
+      "loss": 1.2481,
+      "step": 19775
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.149845126513536e-05,
+      "loss": 2.3209,
+      "step": 19780
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.145822438553441e-05,
+      "loss": 2.9454,
+      "step": 19785
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.141799750593347e-05,
+      "loss": 4.0148,
+      "step": 19790
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.137777062633252e-05,
+      "loss": 4.0611,
+      "step": 19795
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.1337543746731567e-05,
+      "loss": 3.9623,
+      "step": 19800
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.129731686713062e-05,
+      "loss": 2.918,
+      "step": 19805
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.125708998752967e-05,
+      "loss": 3.5767,
+      "step": 19810
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.121686310792873e-05,
+      "loss": 3.4847,
+      "step": 19815
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.1176636228327766e-05,
+      "loss": 1.9963,
+      "step": 19820
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.1136409348726824e-05,
+      "loss": 1.0946,
+      "step": 19825
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.110422784504606e-05,
+      "loss": 1.6789,
+      "step": 19830
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.1064000965445115e-05,
+      "loss": 1.926,
+      "step": 19835
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.1023774085844167e-05,
+      "loss": 4.7094,
+      "step": 19840
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.098354720624321e-05,
+      "loss": 4.5109,
+      "step": 19845
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.094332032664226e-05,
+      "loss": 4.4045,
+      "step": 19850
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.0903093447041314e-05,
+      "loss": 3.6195,
+      "step": 19855
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.086286656744037e-05,
+      "loss": 1.4138,
+      "step": 19860
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.082263968783941e-05,
+      "loss": 2.8188,
+      "step": 19865
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.078241280823847e-05,
+      "loss": 3.5405,
+      "step": 19870
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.074218592863752e-05,
+      "loss": 2.9546,
+      "step": 19875
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.070195904903657e-05,
+      "loss": 3.7135,
+      "step": 19880
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.0661732169435616e-05,
+      "loss": 3.4582,
+      "step": 19885
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.062150528983467e-05,
+      "loss": 4.2113,
+      "step": 19890
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.0581278410233726e-05,
+      "loss": 4.4158,
+      "step": 19895
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.0541051530632764e-05,
+      "loss": 3.453,
+      "step": 19900
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.050082465103182e-05,
+      "loss": 3.6619,
+      "step": 19905
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.0460597771430874e-05,
+      "loss": 3.9289,
+      "step": 19910
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.0420370891829925e-05,
+      "loss": 2.6809,
+      "step": 19915
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.038014401222897e-05,
+      "loss": 0.5294,
+      "step": 19920
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.033991713262802e-05,
+      "loss": 0.7334,
+      "step": 19925
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.029969025302708e-05,
+      "loss": 3.913,
+      "step": 19930
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.025946337342613e-05,
+      "loss": 0.9632,
+      "step": 19935
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.0219236493825176e-05,
+      "loss": 4.6021,
+      "step": 19940
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.017900961422423e-05,
+      "loss": 4.8836,
+      "step": 19945
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.013878273462328e-05,
+      "loss": 3.9064,
+      "step": 19950
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.009855585502233e-05,
+      "loss": 3.9792,
+      "step": 19955
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.0058328975421375e-05,
+      "loss": 4.1586,
+      "step": 19960
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.0018102095820426e-05,
+      "loss": 3.8375,
+      "step": 19965
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.9977875216219484e-05,
+      "loss": 2.0574,
+      "step": 19970
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.993764833661853e-05,
+      "loss": 1.9708,
+      "step": 19975
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.989742145701758e-05,
+      "loss": 1.9487,
+      "step": 19980
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.985719457741663e-05,
+      "loss": 3.1091,
+      "step": 19985
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.9816967697815684e-05,
+      "loss": 5.8838,
+      "step": 19990
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.9776740818214735e-05,
+      "loss": 4.9205,
+      "step": 19995
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.973651393861378e-05,
+      "loss": 4.5146,
+      "step": 20000
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.969628705901284e-05,
+      "loss": 3.2711,
+      "step": 20005
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.965606017941188e-05,
+      "loss": 3.8259,
+      "step": 20010
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.961583329981094e-05,
+      "loss": 3.1064,
+      "step": 20015
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.9575606420209986e-05,
+      "loss": 3.5432,
+      "step": 20020
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.953537954060904e-05,
+      "loss": 1.2155,
+      "step": 20025
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.949515266100809e-05,
+      "loss": 2.7059,
+      "step": 20030
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.945492578140714e-05,
+      "loss": 1.1075,
+      "step": 20035
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.941469890180619e-05,
+      "loss": 3.995,
+      "step": 20040
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.937447202220524e-05,
+      "loss": 4.1667,
+      "step": 20045
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.933424514260429e-05,
+      "loss": 3.6951,
+      "step": 20050
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.929401826300334e-05,
+      "loss": 3.6781,
+      "step": 20055
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.925379138340239e-05,
+      "loss": 4.773,
+      "step": 20060
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.921356450380144e-05,
+      "loss": 3.6868,
+      "step": 20065
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.9173337624200493e-05,
+      "loss": 2.524,
+      "step": 20070
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.9133110744599545e-05,
+      "loss": 1.7292,
+      "step": 20075
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.9092883864998596e-05,
+      "loss": 0.7888,
+      "step": 20080
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.905265698539764e-05,
+      "loss": 2.0986,
+      "step": 20085
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.90124301057967e-05,
+      "loss": 3.924,
+      "step": 20090
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.8972203226195744e-05,
+      "loss": 4.8615,
+      "step": 20095
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.89319763465948e-05,
+      "loss": 3.5187,
+      "step": 20100
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.889174946699385e-05,
+      "loss": 3.4105,
+      "step": 20105
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.88515225873929e-05,
+      "loss": 2.8189,
+      "step": 20110
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.881129570779195e-05,
+      "loss": 3.0407,
+      "step": 20115
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.8771068828191e-05,
+      "loss": 1.8893,
+      "step": 20120
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.873084194859005e-05,
+      "loss": 3.3274,
+      "step": 20125
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.86906150689891e-05,
+      "loss": 2.7007,
+      "step": 20130
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.865038818938815e-05,
+      "loss": 2.7956,
+      "step": 20135
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.86101613097872e-05,
+      "loss": 5.4281,
+      "step": 20140
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.856993443018625e-05,
+      "loss": 4.4959,
+      "step": 20145
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.85297075505853e-05,
+      "loss": 4.1445,
+      "step": 20150
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.8489480670984355e-05,
+      "loss": 3.5901,
+      "step": 20155
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.8449253791383406e-05,
+      "loss": 2.736,
+      "step": 20160
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.840902691178246e-05,
+      "loss": 2.9837,
+      "step": 20165
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.83688000321815e-05,
+      "loss": 2.1882,
+      "step": 20170
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.832857315258056e-05,
+      "loss": 2.8191,
+      "step": 20175
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.8288346272979605e-05,
+      "loss": 1.4353,
+      "step": 20180
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.824811939337866e-05,
+      "loss": 1.7028,
+      "step": 20185
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.820789251377771e-05,
+      "loss": 3.8006,
+      "step": 20190
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.816766563417676e-05,
+      "loss": 3.9104,
+      "step": 20195
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.812743875457581e-05,
+      "loss": 3.7228,
+      "step": 20200
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.8087211874974856e-05,
+      "loss": 3.077,
+      "step": 20205
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.8046984995373914e-05,
+      "loss": 4.0723,
+      "step": 20210
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.800675811577296e-05,
+      "loss": 2.538,
+      "step": 20215
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.796653123617201e-05,
+      "loss": 2.944,
+      "step": 20220
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.792630435657106e-05,
+      "loss": 1.8839,
+      "step": 20225
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.788607747697011e-05,
+      "loss": 2.1462,
+      "step": 20230
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.7845850597369165e-05,
+      "loss": 3.1881,
+      "step": 20235
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.7805623717768216e-05,
+      "loss": 3.8365,
+      "step": 20240
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.776539683816727e-05,
+      "loss": 4.1854,
+      "step": 20245
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.772516995856632e-05,
+      "loss": 3.5527,
+      "step": 20250
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.7684943078965364e-05,
+      "loss": 2.907,
+      "step": 20255
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.764471619936442e-05,
+      "loss": 4.211,
+      "step": 20260
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.760448931976347e-05,
+      "loss": 3.4987,
+      "step": 20265
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.756426244016252e-05,
+      "loss": 2.5673,
+      "step": 20270
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.752403556056157e-05,
+      "loss": 2.0378,
+      "step": 20275
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.7483808680960614e-05,
+      "loss": 2.3546,
+      "step": 20280
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.744358180135967e-05,
+      "loss": 0.8048,
+      "step": 20285
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.740335492175872e-05,
+      "loss": 4.9102,
+      "step": 20290
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.7363128042157776e-05,
+      "loss": 5.3023,
+      "step": 20295
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.732290116255682e-05,
+      "loss": 4.1691,
+      "step": 20300
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.728267428295587e-05,
+      "loss": 3.5041,
+      "step": 20305
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.724244740335492e-05,
+      "loss": 2.0665,
+      "step": 20310
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.7202220523753975e-05,
+      "loss": 3.1756,
+      "step": 20315
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.7161993644153026e-05,
+      "loss": 1.7097,
+      "step": 20320
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.712176676455208e-05,
+      "loss": 2.9661,
+      "step": 20325
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.708153988495113e-05,
+      "loss": 2.5157,
+      "step": 20330
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.704131300535018e-05,
+      "loss": 3.5753,
+      "step": 20335
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.7001086125749225e-05,
+      "loss": 4.2086,
+      "step": 20340
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.6960859246148283e-05,
+      "loss": 4.2891,
+      "step": 20345
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.692063236654733e-05,
+      "loss": 4.4199,
+      "step": 20350
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.688040548694638e-05,
+      "loss": 4.1923,
+      "step": 20355
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.684017860734543e-05,
+      "loss": 1.1563,
+      "step": 20360
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.6799951727744476e-05,
+      "loss": 2.6206,
+      "step": 20365
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.6759724848143534e-05,
+      "loss": 2.1299,
+      "step": 20370
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.671949796854258e-05,
+      "loss": 1.8058,
+      "step": 20375
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.667927108894164e-05,
+      "loss": 2.8761,
+      "step": 20380
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.663904420934068e-05,
+      "loss": 2.907,
+      "step": 20385
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.659881732973973e-05,
+      "loss": 3.6823,
+      "step": 20390
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.6558590450138785e-05,
+      "loss": 3.6055,
+      "step": 20395
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.6518363570537836e-05,
+      "loss": 3.5477,
+      "step": 20400
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.647813669093689e-05,
+      "loss": 3.2163,
+      "step": 20405
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.643790981133594e-05,
+      "loss": 3.3979,
+      "step": 20410
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.6397682931734984e-05,
+      "loss": 3.7832,
+      "step": 20415
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.635745605213404e-05,
+      "loss": 2.9825,
+      "step": 20420
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.631722917253309e-05,
+      "loss": 2.2435,
+      "step": 20425
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.627700229293214e-05,
+      "loss": 1.402,
+      "step": 20430
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.623677541333119e-05,
+      "loss": 1.3045,
+      "step": 20435
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.619654853373024e-05,
+      "loss": 4.5785,
+      "step": 20440
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.615632165412929e-05,
+      "loss": 5.1027,
+      "step": 20445
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.611609477452834e-05,
+      "loss": 3.85,
+      "step": 20450
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.6075867894927395e-05,
+      "loss": 3.7928,
+      "step": 20455
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.603564101532644e-05,
+      "loss": 3.9852,
+      "step": 20460
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.59954141357255e-05,
+      "loss": 3.3365,
+      "step": 20465
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.595518725612454e-05,
+      "loss": 2.475,
+      "step": 20470
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.5914960376523595e-05,
+      "loss": 3.7325,
+      "step": 20475
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.5874733496922646e-05,
+      "loss": 2.4769,
+      "step": 20480
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.58345066173217e-05,
+      "loss": 3.2034,
+      "step": 20485
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.579427973772075e-05,
+      "loss": 4.4615,
+      "step": 20490
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.57540528581198e-05,
+      "loss": 3.6992,
+      "step": 20495
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.5713825978518845e-05,
+      "loss": 3.141,
+      "step": 20500
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.56735990989179e-05,
+      "loss": 3.8582,
+      "step": 20505
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.563337221931695e-05,
+      "loss": 4.3367,
+      "step": 20510
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.5593145339716e-05,
+      "loss": 4.4576,
+      "step": 20515
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.555291846011505e-05,
+      "loss": 0.9217,
+      "step": 20520
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.55126915805141e-05,
+      "loss": 2.4517,
+      "step": 20525
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.5472464700913154e-05,
+      "loss": 0.8062,
+      "step": 20530
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.54322378213122e-05,
+      "loss": 0.7989,
+      "step": 20535
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.539201094171126e-05,
+      "loss": 3.5436,
+      "step": 20540
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.53517840621103e-05,
+      "loss": 3.1012,
+      "step": 20545
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.531155718250936e-05,
+      "loss": 3.8428,
+      "step": 20550
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.5271330302908404e-05,
+      "loss": 3.9062,
+      "step": 20555
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.5231103423307456e-05,
+      "loss": 2.7281,
+      "step": 20560
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.519087654370651e-05,
+      "loss": 3.4412,
+      "step": 20565
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.515064966410556e-05,
+      "loss": 1.8876,
+      "step": 20570
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.511042278450461e-05,
+      "loss": 1.8162,
+      "step": 20575
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.507019590490366e-05,
+      "loss": 3.2144,
+      "step": 20580
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.5029969025302706e-05,
+      "loss": 1.301,
+      "step": 20585
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.498974214570176e-05,
+      "loss": 4.8656,
+      "step": 20590
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.494951526610081e-05,
+      "loss": 4.2117,
+      "step": 20595
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.490928838649986e-05,
+      "loss": 4.4994,
+      "step": 20600
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.486906150689891e-05,
+      "loss": 2.6848,
+      "step": 20605
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.4828834627297964e-05,
+      "loss": 3.4116,
+      "step": 20610
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.4788607747697015e-05,
+      "loss": 3.0542,
+      "step": 20615
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.474838086809606e-05,
+      "loss": 3.0172,
+      "step": 20620
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.470815398849512e-05,
+      "loss": 1.6656,
+      "step": 20625
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.466792710889416e-05,
+      "loss": 2.6862,
+      "step": 20630
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.462770022929322e-05,
+      "loss": 2.2119,
+      "step": 20635
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.4587473349692266e-05,
+      "loss": 5.6586,
+      "step": 20640
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.454724647009132e-05,
+      "loss": 3.9809,
+      "step": 20645
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.450701959049037e-05,
+      "loss": 4.7715,
+      "step": 20650
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.446679271088942e-05,
+      "loss": 4.0703,
+      "step": 20655
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.442656583128847e-05,
+      "loss": 3.8097,
+      "step": 20660
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.4386338951687516e-05,
+      "loss": 3.4632,
+      "step": 20665
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.434611207208657e-05,
+      "loss": 3.296,
+      "step": 20670
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.430588519248562e-05,
+      "loss": 3.2801,
+      "step": 20675
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.426565831288467e-05,
+      "loss": 2.3163,
+      "step": 20680
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.422543143328372e-05,
+      "loss": 4.8768,
+      "step": 20685
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.4185204553682774e-05,
+      "loss": 4.2062,
+      "step": 20690
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.4144977674081825e-05,
+      "loss": 3.9586,
+      "step": 20695
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.410475079448088e-05,
+      "loss": 4.7627,
+      "step": 20700
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.406452391487992e-05,
+      "loss": 4.001,
+      "step": 20705
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.402429703527898e-05,
+      "loss": 3.7379,
+      "step": 20710
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.3984070155678024e-05,
+      "loss": 4.5615,
+      "step": 20715
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.3943843276077076e-05,
+      "loss": 3.2771,
+      "step": 20720
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.390361639647613e-05,
+      "loss": 2.4874,
+      "step": 20725
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.386338951687518e-05,
+      "loss": 1.7379,
+      "step": 20730
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.382316263727423e-05,
+      "loss": 1.266,
+      "step": 20735
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.3782935757673275e-05,
+      "loss": 4.9725,
+      "step": 20740
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.374270887807233e-05,
+      "loss": 4.191,
+      "step": 20745
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.370248199847138e-05,
+      "loss": 3.8467,
+      "step": 20750
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.366225511887043e-05,
+      "loss": 3.1284,
+      "step": 20755
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.362202823926948e-05,
+      "loss": 3.0203,
+      "step": 20760
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.358180135966853e-05,
+      "loss": 3.1714,
+      "step": 20765
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.3541574480067584e-05,
+      "loss": 1.5033,
+      "step": 20770
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.3501347600466635e-05,
+      "loss": 2.4259,
+      "step": 20775
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.3461120720865687e-05,
+      "loss": 3.8117,
+      "step": 20780
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.342089384126474e-05,
+      "loss": 1.745,
+      "step": 20785
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.338066696166378e-05,
+      "loss": 4.2043,
+      "step": 20790
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.334044008206284e-05,
+      "loss": 2.6391,
+      "step": 20795
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.3300213202461886e-05,
+      "loss": 3.6945,
+      "step": 20800
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.325998632286094e-05,
+      "loss": 4.534,
+      "step": 20805
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.321975944325999e-05,
+      "loss": 2.93,
+      "step": 20810
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.317953256365903e-05,
+      "loss": 4.0686,
+      "step": 20815
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.313930568405809e-05,
+      "loss": 4.2174,
+      "step": 20820
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.3099078804457136e-05,
+      "loss": 2.6016,
+      "step": 20825
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.3058851924856194e-05,
+      "loss": 2.1726,
+      "step": 20830
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.301862504525524e-05,
+      "loss": 2.44,
+      "step": 20835
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.297839816565429e-05,
+      "loss": 3.3509,
+      "step": 20840
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.293817128605334e-05,
+      "loss": 3.4732,
+      "step": 20845
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.2897944406452394e-05,
+      "loss": 4.4779,
+      "step": 20850
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.2857717526851445e-05,
+      "loss": 3.4031,
+      "step": 20855
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.2817490647250497e-05,
+      "loss": 2.8928,
+      "step": 20860
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.277726376764955e-05,
+      "loss": 2.8609,
+      "step": 20865
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.27370368880486e-05,
+      "loss": 2.7903,
+      "step": 20870
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.2696810008447644e-05,
+      "loss": 1.7776,
+      "step": 20875
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.26565831288467e-05,
+      "loss": 2.5144,
+      "step": 20880
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.261635624924575e-05,
+      "loss": 1.6099,
+      "step": 20885
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.25761293696448e-05,
+      "loss": 4.2301,
+      "step": 20890
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.253590249004385e-05,
+      "loss": 3.3306,
+      "step": 20895
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.2495675610442895e-05,
+      "loss": 3.6641,
+      "step": 20900
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.245544873084195e-05,
+      "loss": 3.1027,
+      "step": 20905
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.2415221851241e-05,
+      "loss": 4.1236,
+      "step": 20910
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.2374994971640056e-05,
+      "loss": 1.9495,
+      "step": 20915
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.23347680920391e-05,
+      "loss": 1.7783,
+      "step": 20920
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.229454121243815e-05,
+      "loss": 2.336,
+      "step": 20925
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.2254314332837204e-05,
+      "loss": 2.1791,
+      "step": 20930
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.2214087453236255e-05,
+      "loss": 3.2233,
+      "step": 20935
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.2173860573635306e-05,
+      "loss": 3.8101,
+      "step": 20940
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.213363369403436e-05,
+      "loss": 3.4906,
+      "step": 20945
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.20934068144334e-05,
+      "loss": 5.3133,
+      "step": 20950
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.205317993483246e-05,
+      "loss": 3.0161,
+      "step": 20955
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.2012953055231506e-05,
+      "loss": 3.6072,
+      "step": 20960
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.197272617563056e-05,
+      "loss": 3.5251,
+      "step": 20965
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.193249929602961e-05,
+      "loss": 2.6024,
+      "step": 20970
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.189227241642866e-05,
+      "loss": 2.6017,
+      "step": 20975
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.185204553682771e-05,
+      "loss": 3.2057,
+      "step": 20980
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.1811818657226756e-05,
+      "loss": 1.6699,
+      "step": 20985
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.1771591777625814e-05,
+      "loss": 4.5773,
+      "step": 20990
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.173136489802486e-05,
+      "loss": 4.158,
+      "step": 20995
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.169113801842392e-05,
+      "loss": 5.7055,
+      "step": 21000
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.165091113882296e-05,
+      "loss": 3.018,
+      "step": 21005
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.1610684259222013e-05,
+      "loss": 2.5031,
+      "step": 21010
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.1570457379621065e-05,
+      "loss": 3.1937,
+      "step": 21015
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.1530230500020116e-05,
+      "loss": 1.6752,
+      "step": 21020
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.149000362041917e-05,
+      "loss": 1.4042,
+      "step": 21025
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.144977674081822e-05,
+      "loss": 2.4617,
+      "step": 21030
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.1409549861217264e-05,
+      "loss": 3.0018,
+      "step": 21035
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.1369322981616315e-05,
+      "loss": 4.4467,
+      "step": 21040
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.132909610201537e-05,
+      "loss": 4.9004,
+      "step": 21045
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.128886922241442e-05,
+      "loss": 4.0383,
+      "step": 21050
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.124864234281347e-05,
+      "loss": 4.24,
+      "step": 21055
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.120841546321252e-05,
+      "loss": 3.2634,
+      "step": 21060
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.116818858361157e-05,
+      "loss": 2.5876,
+      "step": 21065
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.112796170401062e-05,
+      "loss": 4.6117,
+      "step": 21070
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.1087734824409676e-05,
+      "loss": 0.7939,
+      "step": 21075
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.104750794480872e-05,
+      "loss": 2.6782,
+      "step": 21080
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.100728106520778e-05,
+      "loss": 3.213,
+      "step": 21085
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.096705418560682e-05,
+      "loss": 5.3123,
+      "step": 21090
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.0926827306005875e-05,
+      "loss": 3.9619,
+      "step": 21095
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.0886600426404926e-05,
+      "loss": 4.3588,
+      "step": 21100
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.084637354680398e-05,
+      "loss": 3.2867,
+      "step": 21105
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.080614666720303e-05,
+      "loss": 2.582,
+      "step": 21110
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.076591978760208e-05,
+      "loss": 1.9699,
+      "step": 21115
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.0725692908001125e-05,
+      "loss": 2.4965,
+      "step": 21120
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.068546602840018e-05,
+      "loss": 3.1243,
+      "step": 21125
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.064523914879923e-05,
+      "loss": 1.0647,
+      "step": 21130
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.060501226919828e-05,
+      "loss": 2.6103,
+      "step": 21135
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.056478538959733e-05,
+      "loss": 3.5543,
+      "step": 21140
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.052455850999638e-05,
+      "loss": 4.9754,
+      "step": 21145
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.0484331630395434e-05,
+      "loss": 3.5666,
+      "step": 21150
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.044410475079448e-05,
+      "loss": 3.3064,
+      "step": 21155
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.040387787119354e-05,
+      "loss": 3.4265,
+      "step": 21160
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.036365099159258e-05,
+      "loss": 2.6561,
+      "step": 21165
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.032342411199164e-05,
+      "loss": 1.9494,
+      "step": 21170
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.0283197232390685e-05,
+      "loss": 1.829,
+      "step": 21175
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.0242970352789736e-05,
+      "loss": 2.8679,
+      "step": 21180
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.020274347318879e-05,
+      "loss": 2.5059,
+      "step": 21185
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.016251659358784e-05,
+      "loss": 3.6293,
+      "step": 21190
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.012228971398689e-05,
+      "loss": 4.1984,
+      "step": 21195
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.0082062834385935e-05,
+      "loss": 3.8107,
+      "step": 21200
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.004183595478499e-05,
+      "loss": 2.8273,
+      "step": 21205
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.000160907518404e-05,
+      "loss": 3.2559,
+      "step": 21210
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.996138219558309e-05,
+      "loss": 2.5233,
+      "step": 21215
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.992115531598214e-05,
+      "loss": 2.903,
+      "step": 21220
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.988092843638119e-05,
+      "loss": 3.3081,
+      "step": 21225
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.9840701556780244e-05,
+      "loss": 3.3352,
+      "step": 21230
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.9800474677179296e-05,
+      "loss": 1.8759,
+      "step": 21235
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.976024779757834e-05,
+      "loss": 5.8707,
+      "step": 21240
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.97200209179774e-05,
+      "loss": 4.9869,
+      "step": 21245
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.967979403837644e-05,
+      "loss": 4.4971,
+      "step": 21250
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.9639567158775495e-05,
+      "loss": 3.6207,
+      "step": 21255
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.9599340279174546e-05,
+      "loss": 2.1977,
+      "step": 21260
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.95591133995736e-05,
+      "loss": 2.8388,
+      "step": 21265
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.951888651997265e-05,
+      "loss": 1.8477,
+      "step": 21270
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.9478659640371694e-05,
+      "loss": 1.4987,
+      "step": 21275
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.943843276077075e-05,
+      "loss": 1.4276,
+      "step": 21280
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.93982058811698e-05,
+      "loss": 1.3133,
+      "step": 21285
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.935797900156885e-05,
+      "loss": 4.0492,
+      "step": 21290
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.93177521219679e-05,
+      "loss": 4.2029,
+      "step": 21295
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.927752524236695e-05,
+      "loss": 3.3377,
+      "step": 21300
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.9237298362766e-05,
+      "loss": 4.1035,
+      "step": 21305
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.9197071483165054e-05,
+      "loss": 4.1178,
+      "step": 21310
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.9156844603564105e-05,
+      "loss": 2.9779,
+      "step": 21315
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.911661772396316e-05,
+      "loss": 1.2614,
+      "step": 21320
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.90763908443622e-05,
+      "loss": 1.6433,
+      "step": 21325
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.903616396476126e-05,
+      "loss": 1.6593,
+      "step": 21330
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.8995937085160305e-05,
+      "loss": 4.43,
+      "step": 21335
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.8955710205559356e-05,
+      "loss": 5.0719,
+      "step": 21340
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.891548332595841e-05,
+      "loss": 3.8008,
+      "step": 21345
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.887525644635745e-05,
+      "loss": 4.433,
+      "step": 21350
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.883502956675651e-05,
+      "loss": 3.3838,
+      "step": 21355
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.8794802687155555e-05,
+      "loss": 3.7695,
+      "step": 21360
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.875457580755461e-05,
+      "loss": 2.0561,
+      "step": 21365
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.871434892795366e-05,
+      "loss": 2.6766,
+      "step": 21370
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.867412204835271e-05,
+      "loss": 2.1337,
+      "step": 21375
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.863389516875176e-05,
+      "loss": 2.6355,
+      "step": 21380
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.859366828915081e-05,
+      "loss": 1.5269,
+      "step": 21385
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.8553441409549864e-05,
+      "loss": 4.2592,
+      "step": 21390
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.8513214529948915e-05,
+      "loss": 4.6906,
+      "step": 21395
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.847298765034797e-05,
+      "loss": 3.7479,
+      "step": 21400
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.843276077074702e-05,
+      "loss": 4.2096,
+      "step": 21405
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.839253389114606e-05,
+      "loss": 3.7812,
+      "step": 21410
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.835230701154512e-05,
+      "loss": 5.1412,
+      "step": 21415
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.8312080131944166e-05,
+      "loss": 2.5208,
+      "step": 21420
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.827185325234322e-05,
+      "loss": 1.2869,
+      "step": 21425
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.823162637274227e-05,
+      "loss": 1.4612,
+      "step": 21430
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.8191399493141314e-05,
+      "loss": 4.1741,
+      "step": 21435
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.815117261354037e-05,
+      "loss": 4.3588,
+      "step": 21440
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.8110945733939417e-05,
+      "loss": 3.8693,
+      "step": 21445
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.8070718854338475e-05,
+      "loss": 3.3549,
+      "step": 21450
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.803049197473752e-05,
+      "loss": 3.8469,
+      "step": 21455
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.799026509513657e-05,
+      "loss": 2.6488,
+      "step": 21460
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.795003821553562e-05,
+      "loss": 4.3189,
+      "step": 21465
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.7909811335934674e-05,
+      "loss": 2.1022,
+      "step": 21470
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.7869584456333725e-05,
+      "loss": 3.1552,
+      "step": 21475
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.782935757673278e-05,
+      "loss": 0.6143,
+      "step": 21480
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.778913069713182e-05,
+      "loss": 2.1666,
+      "step": 21485
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.774890381753088e-05,
+      "loss": 4.232,
+      "step": 21490
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.7708676937929924e-05,
+      "loss": 4.3281,
+      "step": 21495
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.7668450058328976e-05,
+      "loss": 4.859,
+      "step": 21500
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.762822317872803e-05,
+      "loss": 3.485,
+      "step": 21505
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.758799629912708e-05,
+      "loss": 3.0216,
+      "step": 21510
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.754776941952613e-05,
+      "loss": 2.9945,
+      "step": 21515
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.7507542539925175e-05,
+      "loss": 3.9133,
+      "step": 21520
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.746731566032423e-05,
+      "loss": 1.9285,
+      "step": 21525
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.742708878072328e-05,
+      "loss": 2.2346,
+      "step": 21530
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.7386861901122336e-05,
+      "loss": 2.376,
+      "step": 21535
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.734663502152138e-05,
+      "loss": 5.018,
+      "step": 21540
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.730640814192043e-05,
+      "loss": 3.4057,
+      "step": 21545
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.7266181262319484e-05,
+      "loss": 4.6818,
+      "step": 21550
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.7225954382718535e-05,
+      "loss": 3.9047,
+      "step": 21555
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.718572750311759e-05,
+      "loss": 3.6093,
+      "step": 21560
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.714550062351664e-05,
+      "loss": 4.4719,
+      "step": 21565
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.710527374391568e-05,
+      "loss": 3.2473,
+      "step": 21570
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.7065046864314734e-05,
+      "loss": 2.2506,
+      "step": 21575
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.7024819984713786e-05,
+      "loss": 2.268,
+      "step": 21580
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.698459310511284e-05,
+      "loss": 3.6383,
+      "step": 21585
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.694436622551189e-05,
+      "loss": 4.4379,
+      "step": 21590
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.690413934591094e-05,
+      "loss": 4.248,
+      "step": 21595
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.686391246630999e-05,
+      "loss": 4.4307,
+      "step": 21600
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.6823685586709036e-05,
+      "loss": 3.4584,
+      "step": 21605
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.6783458707108095e-05,
+      "loss": 4.1742,
+      "step": 21610
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.674323182750714e-05,
+      "loss": 5.0411,
+      "step": 21615
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.67030049479062e-05,
+      "loss": 2.3921,
+      "step": 21620
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.666277806830524e-05,
+      "loss": 2.7103,
+      "step": 21625
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.6622551188704294e-05,
+      "loss": 2.1656,
+      "step": 21630
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.6582324309103345e-05,
+      "loss": 1.949,
+      "step": 21635
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.65420974295024e-05,
+      "loss": 2.9959,
+      "step": 21640
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.650187054990145e-05,
+      "loss": 3.8207,
+      "step": 21645
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.646164367030049e-05,
+      "loss": 3.8807,
+      "step": 21650
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.6421416790699544e-05,
+      "loss": 4.1562,
+      "step": 21655
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.6381189911098596e-05,
+      "loss": 3.2987,
+      "step": 21660
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.634096303149765e-05,
+      "loss": 2.7125,
+      "step": 21665
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.63007361518967e-05,
+      "loss": 2.6132,
+      "step": 21670
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.626050927229575e-05,
+      "loss": 2.0622,
+      "step": 21675
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.62202823926948e-05,
+      "loss": 2.7405,
+      "step": 21680
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.618005551309385e-05,
+      "loss": 2.9218,
+      "step": 21685
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.61398286334929e-05,
+      "loss": 4.7357,
+      "step": 21690
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.6099601753891956e-05,
+      "loss": 4.2771,
+      "step": 21695
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.6059374874291e-05,
+      "loss": 4.242,
+      "step": 21700
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.601914799469006e-05,
+      "loss": 4.1357,
+      "step": 21705
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.5978921115089104e-05,
+      "loss": 2.1266,
+      "step": 21710
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.5938694235488155e-05,
+      "loss": 4.3561,
+      "step": 21715
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.5898467355887207e-05,
+      "loss": 2.5493,
+      "step": 21720
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.585824047628626e-05,
+      "loss": 3.3787,
+      "step": 21725
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.581801359668531e-05,
+      "loss": 2.3158,
+      "step": 21730
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.5777786717084354e-05,
+      "loss": 0.401,
+      "step": 21735
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.5737559837483406e-05,
+      "loss": 4.6857,
+      "step": 21740
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.569733295788246e-05,
+      "loss": 4.5549,
+      "step": 21745
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.565710607828151e-05,
+      "loss": 4.2082,
+      "step": 21750
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.561687919868056e-05,
+      "loss": 3.7506,
+      "step": 21755
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.557665231907961e-05,
+      "loss": 3.4496,
+      "step": 21760
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.553642543947866e-05,
+      "loss": 3.6854,
+      "step": 21765
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.5496198559877714e-05,
+      "loss": 2.3951,
+      "step": 21770
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.545597168027676e-05,
+      "loss": 1.496,
+      "step": 21775
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.541574480067582e-05,
+      "loss": 2.3501,
+      "step": 21780
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.537551792107486e-05,
+      "loss": 2.6589,
+      "step": 21785
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.5335291041473914e-05,
+      "loss": 4.2689,
+      "step": 21790
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.5295064161872965e-05,
+      "loss": 3.6809,
+      "step": 21795
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.5254837282272016e-05,
+      "loss": 3.1069,
+      "step": 21800
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.521461040267107e-05,
+      "loss": 2.6169,
+      "step": 21805
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.517438352307011e-05,
+      "loss": 2.609,
+      "step": 21810
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.513415664346917e-05,
+      "loss": 1.233,
+      "step": 21815
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.5093929763868216e-05,
+      "loss": 3.612,
+      "step": 21820
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.505370288426727e-05,
+      "loss": 2.2503,
+      "step": 21825
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.501347600466632e-05,
+      "loss": 2.9724,
+      "step": 21830
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.497324912506537e-05,
+      "loss": 1.8672,
+      "step": 21835
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.493302224546442e-05,
+      "loss": 5.3764,
+      "step": 21840
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.489279536586347e-05,
+      "loss": 4.4464,
+      "step": 21845
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.4852568486262524e-05,
+      "loss": 3.6959,
+      "step": 21850
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.4812341606661576e-05,
+      "loss": 3.6832,
+      "step": 21855
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.477211472706062e-05,
+      "loss": 3.5953,
+      "step": 21860
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.473188784745968e-05,
+      "loss": 3.1239,
+      "step": 21865
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.4691660967858723e-05,
+      "loss": 2.1076,
+      "step": 21870
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.4651434088257775e-05,
+      "loss": 3.3783,
+      "step": 21875
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.4611207208656826e-05,
+      "loss": 1.4034,
+      "step": 21880
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.457098032905587e-05,
+      "loss": 1.7014,
+      "step": 21885
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.453075344945493e-05,
+      "loss": 5.3859,
+      "step": 21890
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.4490526569853974e-05,
+      "loss": 4.724,
+      "step": 21895
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.445029969025303e-05,
+      "loss": 3.8088,
+      "step": 21900
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.441007281065208e-05,
+      "loss": 3.1448,
+      "step": 21905
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.436984593105113e-05,
+      "loss": 3.1462,
+      "step": 21910
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.432961905145018e-05,
+      "loss": 2.371,
+      "step": 21915
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.428939217184923e-05,
+      "loss": 2.8063,
+      "step": 21920
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.424916529224828e-05,
+      "loss": 2.514,
+      "step": 21925
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.4208938412647334e-05,
+      "loss": 3.2729,
+      "step": 21930
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.4168711533046386e-05,
+      "loss": 2.0309,
+      "step": 21935
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.412848465344544e-05,
+      "loss": 3.9336,
+      "step": 21940
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.408825777384448e-05,
+      "loss": 4.7568,
+      "step": 21945
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.404803089424354e-05,
+      "loss": 3.4958,
+      "step": 21950
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.4007804014642585e-05,
+      "loss": 3.6162,
+      "step": 21955
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.3967577135041636e-05,
+      "loss": 3.7423,
+      "step": 21960
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.392735025544069e-05,
+      "loss": 1.3449,
+      "step": 21965
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.388712337583973e-05,
+      "loss": 2.8116,
+      "step": 21970
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.384689649623879e-05,
+      "loss": 2.0983,
+      "step": 21975
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.3806669616637835e-05,
+      "loss": 3.2099,
+      "step": 21980
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.3766442737036894e-05,
+      "loss": 1.4468,
+      "step": 21985
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.372621585743594e-05,
+      "loss": 4.3094,
+      "step": 21990
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.368598897783499e-05,
+      "loss": 2.935,
+      "step": 21995
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.364576209823404e-05,
+      "loss": 4.1764,
+      "step": 22000
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.360553521863309e-05,
+      "loss": 4.1816,
+      "step": 22005
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.3565308339032144e-05,
+      "loss": 2.4688,
+      "step": 22010
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.3525081459431196e-05,
+      "loss": 3.324,
+      "step": 22015
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.348485457983024e-05,
+      "loss": 3.443,
+      "step": 22020
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.34446277002293e-05,
+      "loss": 2.2952,
+      "step": 22025
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.340440082062834e-05,
+      "loss": 2.7904,
+      "step": 22030
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.3364173941027395e-05,
+      "loss": 3.2367,
+      "step": 22035
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.3323947061426446e-05,
+      "loss": 3.6912,
+      "step": 22040
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.32837201818255e-05,
+      "loss": 4.3529,
+      "step": 22045
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.324349330222455e-05,
+      "loss": 4.1754,
+      "step": 22050
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.3203266422623594e-05,
+      "loss": 4.0643,
+      "step": 22055
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.316303954302265e-05,
+      "loss": 3.0192,
+      "step": 22060
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.31228126634217e-05,
+      "loss": 3.0924,
+      "step": 22065
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.3082585783820755e-05,
+      "loss": 2.415,
+      "step": 22070
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.30423589042198e-05,
+      "loss": 3.3243,
+      "step": 22075
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.300213202461885e-05,
+      "loss": 1.2963,
+      "step": 22080
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.29619051450179e-05,
+      "loss": 3.9981,
+      "step": 22085
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.2921678265416954e-05,
+      "loss": 3.3592,
+      "step": 22090
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.2881451385816006e-05,
+      "loss": 3.4289,
+      "step": 22095
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.284122450621506e-05,
+      "loss": 3.7771,
+      "step": 22100
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.28009976266141e-05,
+      "loss": 3.6328,
+      "step": 22105
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.276077074701315e-05,
+      "loss": 3.0618,
+      "step": 22110
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.2720543867412205e-05,
+      "loss": 2.6314,
+      "step": 22115
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.2680316987811256e-05,
+      "loss": 1.7609,
+      "step": 22120
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.264009010821031e-05,
+      "loss": 1.2902,
+      "step": 22125
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.259986322860936e-05,
+      "loss": 0.4134,
+      "step": 22130
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.255963634900841e-05,
+      "loss": 2.865,
+      "step": 22135
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.2519409469407455e-05,
+      "loss": 4.5992,
+      "step": 22140
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.2479182589806514e-05,
+      "loss": 4.1874,
+      "step": 22145
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.243895571020556e-05,
+      "loss": 2.4801,
+      "step": 22150
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.2398728830604616e-05,
+      "loss": 3.9021,
+      "step": 22155
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.235850195100366e-05,
+      "loss": 3.6825,
+      "step": 22160
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.231827507140271e-05,
+      "loss": 3.2391,
+      "step": 22165
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.2278048191801764e-05,
+      "loss": 3.5101,
+      "step": 22170
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.2237821312200816e-05,
+      "loss": 3.7007,
+      "step": 22175
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.219759443259987e-05,
+      "loss": 1.1778,
+      "step": 22180
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.215736755299891e-05,
+      "loss": 0.9388,
+      "step": 22185
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.211714067339796e-05,
+      "loss": 4.8832,
+      "step": 22190
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.2076913793797015e-05,
+      "loss": 3.309,
+      "step": 22195
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.2036686914196066e-05,
+      "loss": 4.3115,
+      "step": 22200
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.199646003459512e-05,
+      "loss": 4.6436,
+      "step": 22205
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.195623315499417e-05,
+      "loss": 3.9236,
+      "step": 22210
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.191600627539322e-05,
+      "loss": 2.4813,
+      "step": 22215
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.187577939579227e-05,
+      "loss": 1.7186,
+      "step": 22220
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.183555251619132e-05,
+      "loss": 3.0956,
+      "step": 22225
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.1795325636590375e-05,
+      "loss": 0.8746,
+      "step": 22230
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.175509875698942e-05,
+      "loss": 2.345,
+      "step": 22235
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.171487187738848e-05,
+      "loss": 3.7502,
+      "step": 22240
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.167464499778752e-05,
+      "loss": 4.0754,
+      "step": 22245
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.1634418118186574e-05,
+      "loss": 5.118,
+      "step": 22250
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.1594191238585625e-05,
+      "loss": 5.1141,
+      "step": 22255
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.155396435898468e-05,
+      "loss": 3.1888,
+      "step": 22260
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.151373747938373e-05,
+      "loss": 2.4457,
+      "step": 22265
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.147351059978277e-05,
+      "loss": 3.7101,
+      "step": 22270
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.1433283720181825e-05,
+      "loss": 2.8647,
+      "step": 22275
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.1393056840580876e-05,
+      "loss": 2.2749,
+      "step": 22280
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.135282996097993e-05,
+      "loss": 2.0576,
+      "step": 22285
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.131260308137898e-05,
+      "loss": 4.5924,
+      "step": 22290
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.127237620177803e-05,
+      "loss": 3.8521,
+      "step": 22295
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.123214932217708e-05,
+      "loss": 3.3519,
+      "step": 22300
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.119192244257613e-05,
+      "loss": 3.1276,
+      "step": 22305
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.115169556297518e-05,
+      "loss": 3.8262,
+      "step": 22310
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.1111468683374236e-05,
+      "loss": 2.2402,
+      "step": 22315
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.107124180377328e-05,
+      "loss": 2.7102,
+      "step": 22320
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.103101492417233e-05,
+      "loss": 1.8521,
+      "step": 22325
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.0990788044571384e-05,
+      "loss": 1.5898,
+      "step": 22330
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.0950561164970435e-05,
+      "loss": 3.3699,
+      "step": 22335
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.091033428536949e-05,
+      "loss": 3.958,
+      "step": 22340
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.087010740576853e-05,
+      "loss": 4.5051,
+      "step": 22345
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.082988052616759e-05,
+      "loss": 3.4871,
+      "step": 22350
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.0789653646566634e-05,
+      "loss": 2.7721,
+      "step": 22355
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.0749426766965686e-05,
+      "loss": 3.9631,
+      "step": 22360
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.070919988736474e-05,
+      "loss": 4.0345,
+      "step": 22365
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.066897300776379e-05,
+      "loss": 2.9204,
+      "step": 22370
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.062874612816284e-05,
+      "loss": 2.6203,
+      "step": 22375
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.058851924856189e-05,
+      "loss": 2.0655,
+      "step": 22380
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.054829236896094e-05,
+      "loss": 2.9641,
+      "step": 22385
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.0508065489359995e-05,
+      "loss": 3.8604,
+      "step": 22390
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.0467838609759043e-05,
+      "loss": 4.3584,
+      "step": 22395
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.0427611730158094e-05,
+      "loss": 4.0123,
+      "step": 22400
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.0387384850557142e-05,
+      "loss": 4.0429,
+      "step": 22405
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.0347157970956197e-05,
+      "loss": 3.8104,
+      "step": 22410
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.0306931091355245e-05,
+      "loss": 2.3843,
+      "step": 22415
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.0266704211754293e-05,
+      "loss": 2.1063,
+      "step": 22420
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.0226477332153348e-05,
+      "loss": 2.0419,
+      "step": 22425
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.0186250452552396e-05,
+      "loss": 1.0568,
+      "step": 22430
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.015406894887164e-05,
+      "loss": 3.2819,
+      "step": 22435
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.0113842069270688e-05,
+      "loss": 5.0303,
+      "step": 22440
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.007361518966974e-05,
+      "loss": 3.7967,
+      "step": 22445
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.0033388310068787e-05,
+      "loss": 3.6611,
+      "step": 22450
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.9993161430467842e-05,
+      "loss": 3.6416,
+      "step": 22455
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.995293455086689e-05,
+      "loss": 1.9096,
+      "step": 22460
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.9912707671265945e-05,
+      "loss": 1.793,
+      "step": 22465
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.9872480791664993e-05,
+      "loss": 2.3129,
+      "step": 22470
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.983225391206404e-05,
+      "loss": 0.621,
+      "step": 22475
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.9792027032463093e-05,
+      "loss": 1.0425,
+      "step": 22480
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.975180015286214e-05,
+      "loss": 2.1539,
+      "step": 22485
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.9711573273261196e-05,
+      "loss": 4.5879,
+      "step": 22490
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.9671346393660244e-05,
+      "loss": 5.5826,
+      "step": 22495
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.96311195140593e-05,
+      "loss": 4.015,
+      "step": 22500
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.9590892634458343e-05,
+      "loss": 3.9393,
+      "step": 22505
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.9550665754857398e-05,
+      "loss": 4.0424,
+      "step": 22510
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.9510438875256446e-05,
+      "loss": 2.3775,
+      "step": 22515
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.94702119956555e-05,
+      "loss": 3.9154,
+      "step": 22520
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.942998511605455e-05,
+      "loss": 3.6949,
+      "step": 22525
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.93897582364536e-05,
+      "loss": 3.6259,
+      "step": 22530
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.934953135685265e-05,
+      "loss": 2.0479,
+      "step": 22535
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.9309304477251703e-05,
+      "loss": 4.7387,
+      "step": 22540
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.926907759765075e-05,
+      "loss": 4.1223,
+      "step": 22545
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.9228850718049806e-05,
+      "loss": 3.8104,
+      "step": 22550
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.9188623838448854e-05,
+      "loss": 2.8752,
+      "step": 22555
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.9148396958847902e-05,
+      "loss": 3.5936,
+      "step": 22560
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.9108170079246954e-05,
+      "loss": 3.1607,
+      "step": 22565
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.9067943199646002e-05,
+      "loss": 2.8457,
+      "step": 22570
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.9027716320045057e-05,
+      "loss": 2.4739,
+      "step": 22575
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.8987489440444105e-05,
+      "loss": 3.7004,
+      "step": 22580
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.8947262560843156e-05,
+      "loss": 2.4516,
+      "step": 22585
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.8907035681242205e-05,
+      "loss": 4.2662,
+      "step": 22590
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.886680880164126e-05,
+      "loss": 3.1941,
+      "step": 22595
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.8826581922040307e-05,
+      "loss": 4.3604,
+      "step": 22600
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.8786355042439362e-05,
+      "loss": 4.5975,
+      "step": 22605
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.874612816283841e-05,
+      "loss": 2.4618,
+      "step": 22610
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.8705901283237462e-05,
+      "loss": 2.5024,
+      "step": 22615
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.866567440363651e-05,
+      "loss": 2.0588,
+      "step": 22620
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.8625447524035565e-05,
+      "loss": 2.6591,
+      "step": 22625
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.8585220644434613e-05,
+      "loss": 1.1996,
+      "step": 22630
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.854499376483366e-05,
+      "loss": 1.5975,
+      "step": 22635
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.8504766885232716e-05,
+      "loss": 4.0307,
+      "step": 22640
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.8464540005631764e-05,
+      "loss": 4.334,
+      "step": 22645
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.8424313126030815e-05,
+      "loss": 3.3168,
+      "step": 22650
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.8384086246429863e-05,
+      "loss": 3.9016,
+      "step": 22655
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.8343859366828918e-05,
+      "loss": 2.4265,
+      "step": 22660
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.8303632487227966e-05,
+      "loss": 2.7111,
+      "step": 22665
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.8263405607627018e-05,
+      "loss": 2.3812,
+      "step": 22670
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.8223178728026066e-05,
+      "loss": 2.0577,
+      "step": 22675
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.818295184842512e-05,
+      "loss": 1.1668,
+      "step": 22680
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.814272496882417e-05,
+      "loss": 2.8699,
+      "step": 22685
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.8102498089223224e-05,
+      "loss": 4.2182,
+      "step": 22690
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.8062271209622272e-05,
+      "loss": 4.5916,
+      "step": 22695
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.8022044330021323e-05,
+      "loss": 4.7451,
+      "step": 22700
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.798181745042037e-05,
+      "loss": 4.1365,
+      "step": 22705
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.794159057081942e-05,
+      "loss": 2.2262,
+      "step": 22710
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.7901363691218474e-05,
+      "loss": 2.2605,
+      "step": 22715
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.7861136811617522e-05,
+      "loss": 3.7399,
+      "step": 22720
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.7820909932016577e-05,
+      "loss": 2.6796,
+      "step": 22725
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.7780683052415625e-05,
+      "loss": 2.1081,
+      "step": 22730
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.7740456172814677e-05,
+      "loss": 3.9569,
+      "step": 22735
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.7700229293213725e-05,
+      "loss": 5.2086,
+      "step": 22740
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.766000241361278e-05,
+      "loss": 4.1189,
+      "step": 22745
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.7619775534011828e-05,
+      "loss": 3.4363,
+      "step": 22750
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.757954865441088e-05,
+      "loss": 4.1549,
+      "step": 22755
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.7539321774809927e-05,
+      "loss": 3.4964,
+      "step": 22760
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.7499094895208982e-05,
+      "loss": 3.3692,
+      "step": 22765
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.745886801560803e-05,
+      "loss": 1.6907,
+      "step": 22770
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.7418641136007085e-05,
+      "loss": 1.4017,
+      "step": 22775
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.7378414256406133e-05,
+      "loss": 1.4608,
+      "step": 22780
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.733818737680518e-05,
+      "loss": 2.8462,
+      "step": 22785
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.7297960497204233e-05,
+      "loss": 4.6463,
+      "step": 22790
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.725773361760328e-05,
+      "loss": 4.1804,
+      "step": 22795
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.7217506738002336e-05,
+      "loss": 3.4373,
+      "step": 22800
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.7177279858401384e-05,
+      "loss": 2.9582,
+      "step": 22805
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.7137052978800435e-05,
+      "loss": 3.3782,
+      "step": 22810
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.7096826099199483e-05,
+      "loss": 3.1693,
+      "step": 22815
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.7056599219598538e-05,
+      "loss": 2.3896,
+      "step": 22820
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.7016372339997586e-05,
+      "loss": 2.9684,
+      "step": 22825
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.697614546039664e-05,
+      "loss": 2.6333,
+      "step": 22830
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.693591858079569e-05,
+      "loss": 4.8025,
+      "step": 22835
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.689569170119474e-05,
+      "loss": 3.923,
+      "step": 22840
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.685546482159379e-05,
+      "loss": 3.8394,
+      "step": 22845
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.6815237941992844e-05,
+      "loss": 4.2166,
+      "step": 22850
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.677501106239189e-05,
+      "loss": 3.2895,
+      "step": 22855
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.673478418279094e-05,
+      "loss": 3.2745,
+      "step": 22860
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.6694557303189995e-05,
+      "loss": 2.9658,
+      "step": 22865
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.6654330423589043e-05,
+      "loss": 2.6981,
+      "step": 22870
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.6614103543988094e-05,
+      "loss": 1.1102,
+      "step": 22875
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.6573876664387142e-05,
+      "loss": 0.2825,
+      "step": 22880
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.6533649784786197e-05,
+      "loss": 0.6763,
+      "step": 22885
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.6493422905185245e-05,
+      "loss": 3.8113,
+      "step": 22890
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.6453196025584297e-05,
+      "loss": 3.5855,
+      "step": 22895
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.6412969145983345e-05,
+      "loss": 4.2107,
+      "step": 22900
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.63727422663824e-05,
+      "loss": 3.6761,
+      "step": 22905
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.6332515386781448e-05,
+      "loss": 2.6449,
+      "step": 22910
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.6292288507180502e-05,
+      "loss": 1.46,
+      "step": 22915
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.625206162757955e-05,
+      "loss": 2.5006,
+      "step": 22920
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.6211834747978602e-05,
+      "loss": 1.2944,
+      "step": 22925
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.617160786837765e-05,
+      "loss": 3.3048,
+      "step": 22930
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.6131380988776698e-05,
+      "loss": 2.9723,
+      "step": 22935
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.6091154109175753e-05,
+      "loss": 4.3123,
+      "step": 22940
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.60509272295748e-05,
+      "loss": 5.226,
+      "step": 22945
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.6010700349973856e-05,
+      "loss": 3.9012,
+      "step": 22950
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.5970473470372904e-05,
+      "loss": 3.7363,
+      "step": 22955
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.5930246590771956e-05,
+      "loss": 3.6529,
+      "step": 22960
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.5890019711171004e-05,
+      "loss": 2.3566,
+      "step": 22965
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.584979283157006e-05,
+      "loss": 2.1602,
+      "step": 22970
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.5809565951969107e-05,
+      "loss": 3.2619,
+      "step": 22975
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.5769339072368158e-05,
+      "loss": 2.565,
+      "step": 22980
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.5729112192767206e-05,
+      "loss": 3.3445,
+      "step": 22985
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.568888531316626e-05,
+      "loss": 4.1391,
+      "step": 22990
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.564865843356531e-05,
+      "loss": 3.7564,
+      "step": 22995
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.5608431553964364e-05,
+      "loss": 3.3184,
+      "step": 23000
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.5568204674363412e-05,
+      "loss": 3.8339,
+      "step": 23005
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.552797779476246e-05,
+      "loss": 2.4366,
+      "step": 23010
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.548775091516151e-05,
+      "loss": 2.4264,
+      "step": 23015
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.544752403556056e-05,
+      "loss": 1.9581,
+      "step": 23020
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.5407297155959614e-05,
+      "loss": 1.901,
+      "step": 23025
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.5367070276358662e-05,
+      "loss": 0.0299,
+      "step": 23030
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.5326843396757717e-05,
+      "loss": 1.0677,
+      "step": 23035
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.5286616517156762e-05,
+      "loss": 4.8,
+      "step": 23040
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.5246389637555817e-05,
+      "loss": 3.8922,
+      "step": 23045
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.5206162757954865e-05,
+      "loss": 3.9426,
+      "step": 23050
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.516593587835392e-05,
+      "loss": 4.1615,
+      "step": 23055
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.5125708998752968e-05,
+      "loss": 4.2786,
+      "step": 23060
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.508548211915202e-05,
+      "loss": 3.0869,
+      "step": 23065
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.5045255239551067e-05,
+      "loss": 2.5565,
+      "step": 23070
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.5005028359950122e-05,
+      "loss": 2.0955,
+      "step": 23075
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.496480148034917e-05,
+      "loss": 1.9791,
+      "step": 23080
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.4924574600748222e-05,
+      "loss": 1.7667,
+      "step": 23085
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.4884347721147273e-05,
+      "loss": 3.9285,
+      "step": 23090
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.484412084154632e-05,
+      "loss": 4.3834,
+      "step": 23095
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.4803893961945373e-05,
+      "loss": 4.3969,
+      "step": 23100
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.4763667082344424e-05,
+      "loss": 4.2646,
+      "step": 23105
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.4723440202743476e-05,
+      "loss": 3.1369,
+      "step": 23110
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.4683213323142527e-05,
+      "loss": 3.584,
+      "step": 23115
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.4642986443541575e-05,
+      "loss": 2.7839,
+      "step": 23120
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.4602759563940623e-05,
+      "loss": 1.9654,
+      "step": 23125
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.4562532684339675e-05,
+      "loss": 2.4053,
+      "step": 23130
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.4522305804738726e-05,
+      "loss": 2.3811,
+      "step": 23135
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.4482078925137778e-05,
+      "loss": 4.3805,
+      "step": 23140
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.444185204553683e-05,
+      "loss": 3.634,
+      "step": 23145
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.440162516593588e-05,
+      "loss": 4.3363,
+      "step": 23150
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.436139828633493e-05,
+      "loss": 4.2129,
+      "step": 23155
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.432117140673398e-05,
+      "loss": 3.1537,
+      "step": 23160
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.4280944527133032e-05,
+      "loss": 3.78,
+      "step": 23165
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.4240717647532083e-05,
+      "loss": 3.6497,
+      "step": 23170
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.4200490767931135e-05,
+      "loss": 4.7099,
+      "step": 23175
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.4160263888330183e-05,
+      "loss": 1.8507,
+      "step": 23180
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.4120037008729234e-05,
+      "loss": 2.9047,
+      "step": 23185
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.4079810129128286e-05,
+      "loss": 4.0793,
+      "step": 23190
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.4039583249527337e-05,
+      "loss": 4.4531,
+      "step": 23195
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.3999356369926385e-05,
+      "loss": 5.0652,
+      "step": 23200
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.3959129490325437e-05,
+      "loss": 3.3764,
+      "step": 23205
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.3918902610724485e-05,
+      "loss": 3.6626,
+      "step": 23210
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.3878675731123536e-05,
+      "loss": 3.4744,
+      "step": 23215
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.3838448851522588e-05,
+      "loss": 2.2833,
+      "step": 23220
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.379822197192164e-05,
+      "loss": 0.5065,
+      "step": 23225
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.375799509232069e-05,
+      "loss": 2.4229,
+      "step": 23230
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.371776821271974e-05,
+      "loss": 0.7311,
+      "step": 23235
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.367754133311879e-05,
+      "loss": 3.8694,
+      "step": 23240
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.3637314453517842e-05,
+      "loss": 3.3383,
+      "step": 23245
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.3597087573916893e-05,
+      "loss": 3.9184,
+      "step": 23250
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.3556860694315945e-05,
+      "loss": 3.0708,
+      "step": 23255
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.3516633814714996e-05,
+      "loss": 2.9025,
+      "step": 23260
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.3476406935114044e-05,
+      "loss": 2.6064,
+      "step": 23265
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.3436180055513096e-05,
+      "loss": 1.7971,
+      "step": 23270
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.3395953175912147e-05,
+      "loss": 2.8858,
+      "step": 23275
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.3355726296311195e-05,
+      "loss": 3.6838,
+      "step": 23280
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.3315499416710247e-05,
+      "loss": 2.2048,
+      "step": 23285
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.3275272537109298e-05,
+      "loss": 4.4943,
+      "step": 23290
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.3235045657508346e-05,
+      "loss": 4.283,
+      "step": 23295
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.3194818777907398e-05,
+      "loss": 4.518,
+      "step": 23300
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.315459189830645e-05,
+      "loss": 3.516,
+      "step": 23305
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.31143650187055e-05,
+      "loss": 4.2672,
+      "step": 23310
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.3074138139104552e-05,
+      "loss": 3.4592,
+      "step": 23315
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.30339112595036e-05,
+      "loss": 3.5898,
+      "step": 23320
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.299368437990265e-05,
+      "loss": 2.7908,
+      "step": 23325
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.2953457500301703e-05,
+      "loss": 2.1385,
+      "step": 23330
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.2913230620700755e-05,
+      "loss": 1.7825,
+      "step": 23335
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.2873003741099806e-05,
+      "loss": 4.9459,
+      "step": 23340
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.2832776861498854e-05,
+      "loss": 3.7873,
+      "step": 23345
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.2792549981897906e-05,
+      "loss": 3.7168,
+      "step": 23350
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.2752323102296954e-05,
+      "loss": 3.9023,
+      "step": 23355
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.2712096222696005e-05,
+      "loss": 1.3564,
+      "step": 23360
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.2671869343095057e-05,
+      "loss": 3.3062,
+      "step": 23365
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.2631642463494108e-05,
+      "loss": 4.416,
+      "step": 23370
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.259141558389316e-05,
+      "loss": 3.2855,
+      "step": 23375
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.2551188704292208e-05,
+      "loss": 3.3284,
+      "step": 23380
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.251096182469126e-05,
+      "loss": 2.627,
+      "step": 23385
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.247073494509031e-05,
+      "loss": 4.008,
+      "step": 23390
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.2430508065489362e-05,
+      "loss": 4.5375,
+      "step": 23395
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.2390281185888413e-05,
+      "loss": 4.3504,
+      "step": 23400
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.235005430628746e-05,
+      "loss": 2.7492,
+      "step": 23405
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.2309827426686513e-05,
+      "loss": 2.9355,
+      "step": 23410
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.2269600547085564e-05,
+      "loss": 2.2093,
+      "step": 23415
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.2229373667484616e-05,
+      "loss": 2.8348,
+      "step": 23420
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.2189146787883667e-05,
+      "loss": 2.7705,
+      "step": 23425
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.2148919908282715e-05,
+      "loss": 1.4777,
+      "step": 23430
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.2108693028681764e-05,
+      "loss": 2.0384,
+      "step": 23435
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.2068466149080815e-05,
+      "loss": 3.5889,
+      "step": 23440
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.2028239269479867e-05,
+      "loss": 4.2697,
+      "step": 23445
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.1988012389878918e-05,
+      "loss": 3.7738,
+      "step": 23450
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.194778551027797e-05,
+      "loss": 3.7055,
+      "step": 23455
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.1907558630677018e-05,
+      "loss": 3.8967,
+      "step": 23460
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.186733175107607e-05,
+      "loss": 4.0676,
+      "step": 23465
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.182710487147512e-05,
+      "loss": 3.5904,
+      "step": 23470
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.1786877991874172e-05,
+      "loss": 1.4659,
+      "step": 23475
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.1746651112273223e-05,
+      "loss": 4.3659,
+      "step": 23480
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.1706424232672275e-05,
+      "loss": 2.4338,
+      "step": 23485
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.1666197353071323e-05,
+      "loss": 3.2225,
+      "step": 23490
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.1625970473470374e-05,
+      "loss": 3.9035,
+      "step": 23495
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.1585743593869426e-05,
+      "loss": 4.0298,
+      "step": 23500
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.1545516714268474e-05,
+      "loss": 4.2254,
+      "step": 23505
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.1505289834667525e-05,
+      "loss": 2.9377,
+      "step": 23510
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.1465062955066577e-05,
+      "loss": 2.3562,
+      "step": 23515
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.1424836075465625e-05,
+      "loss": 1.3816,
+      "step": 23520
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.1384609195864676e-05,
+      "loss": 2.5952,
+      "step": 23525
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.1344382316263728e-05,
+      "loss": 1.156,
+      "step": 23530
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.130415543666278e-05,
+      "loss": 2.085,
+      "step": 23535
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.126392855706183e-05,
+      "loss": 4.2512,
+      "step": 23540
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.122370167746088e-05,
+      "loss": 3.7354,
+      "step": 23545
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.118347479785993e-05,
+      "loss": 3.8537,
+      "step": 23550
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.1143247918258982e-05,
+      "loss": 3.9422,
+      "step": 23555
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.1103021038658033e-05,
+      "loss": 2.9316,
+      "step": 23560
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.1062794159057085e-05,
+      "loss": 3.9269,
+      "step": 23565
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.1022567279456136e-05,
+      "loss": 1.4704,
+      "step": 23570
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0982340399855184e-05,
+      "loss": 2.5099,
+      "step": 23575
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0942113520254236e-05,
+      "loss": 2.5829,
+      "step": 23580
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0901886640653284e-05,
+      "loss": 4.3772,
+      "step": 23585
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0861659761052335e-05,
+      "loss": 3.4992,
+      "step": 23590
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0821432881451387e-05,
+      "loss": 4.1639,
+      "step": 23595
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0781206001850438e-05,
+      "loss": 3.1853,
+      "step": 23600
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0740979122249486e-05,
+      "loss": 3.6627,
+      "step": 23605
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0700752242648538e-05,
+      "loss": 3.7691,
+      "step": 23610
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.066052536304759e-05,
+      "loss": 2.473,
+      "step": 23615
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.062029848344664e-05,
+      "loss": 2.8132,
+      "step": 23620
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.0580071603845692e-05,
+      "loss": 1.6381,
+      "step": 23625
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.053984472424474e-05,
+      "loss": 1.6351,
+      "step": 23630
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.0499617844643792e-05,
+      "loss": 2.8561,
+      "step": 23635
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.0459390965042843e-05,
+      "loss": 3.5916,
+      "step": 23640
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.0419164085441895e-05,
+      "loss": 3.0802,
+      "step": 23645
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.0378937205840946e-05,
+      "loss": 4.2158,
+      "step": 23650
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.0338710326239994e-05,
+      "loss": 1.5458,
+      "step": 23655
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.0298483446639042e-05,
+      "loss": 3.1277,
+      "step": 23660
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.0258256567038094e-05,
+      "loss": 3.5196,
+      "step": 23665
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.0218029687437145e-05,
+      "loss": 3.1044,
+      "step": 23670
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.0177802807836197e-05,
+      "loss": 4.1605,
+      "step": 23675
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.0137575928235248e-05,
+      "loss": 1.0403,
+      "step": 23680
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.00973490486343e-05,
+      "loss": 3.4376,
+      "step": 23685
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.0057122169033348e-05,
+      "loss": 3.7799,
+      "step": 23690
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.00168952894324e-05,
+      "loss": 5.3645,
+      "step": 23695
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.997666840983145e-05,
+      "loss": 4.4109,
+      "step": 23700
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9936441530230502e-05,
+      "loss": 4.6688,
+      "step": 23705
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9896214650629554e-05,
+      "loss": 3.8071,
+      "step": 23710
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.98559877710286e-05,
+      "loss": 3.7666,
+      "step": 23715
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9815760891427653e-05,
+      "loss": 2.0954,
+      "step": 23720
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9775534011826705e-05,
+      "loss": 2.5181,
+      "step": 23725
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9735307132225756e-05,
+      "loss": 3.2477,
+      "step": 23730
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9695080252624804e-05,
+      "loss": 1.3823,
+      "step": 23735
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9654853373023856e-05,
+      "loss": 4.2088,
+      "step": 23740
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9614626493422904e-05,
+      "loss": 4.3371,
+      "step": 23745
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9574399613821955e-05,
+      "loss": 4.924,
+      "step": 23750
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9534172734221007e-05,
+      "loss": 2.5295,
+      "step": 23755
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9493945854620058e-05,
+      "loss": 2.1711,
+      "step": 23760
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.945371897501911e-05,
+      "loss": 1.7804,
+      "step": 23765
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9413492095418158e-05,
+      "loss": 2.7061,
+      "step": 23770
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.937326521581721e-05,
+      "loss": 1.8079,
+      "step": 23775
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.933303833621626e-05,
+      "loss": 2.6042,
+      "step": 23780
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9292811456615312e-05,
+      "loss": 1.4644,
+      "step": 23785
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9252584577014364e-05,
+      "loss": 3.4973,
+      "step": 23790
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9212357697413415e-05,
+      "loss": 5.3607,
+      "step": 23795
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9172130817812463e-05,
+      "loss": 2.8048,
+      "step": 23800
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9131903938211515e-05,
+      "loss": 1.9658,
+      "step": 23805
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9091677058610566e-05,
+      "loss": 3.1126,
+      "step": 23810
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9051450179009614e-05,
+      "loss": 2.9254,
+      "step": 23815
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9011223299408666e-05,
+      "loss": 0.5672,
+      "step": 23820
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.8970996419807717e-05,
+      "loss": 2.4323,
+      "step": 23825
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.8930769540206765e-05,
+      "loss": 1.714,
+      "step": 23830
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.8890542660605817e-05,
+      "loss": 1.6771,
+      "step": 23835
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.8850315781004868e-05,
+      "loss": 5.8141,
+      "step": 23840
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.881008890140392e-05,
+      "loss": 3.5088,
+      "step": 23845
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.876986202180297e-05,
+      "loss": 3.4639,
+      "step": 23850
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.872963514220202e-05,
+      "loss": 4.4504,
+      "step": 23855
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.868940826260107e-05,
+      "loss": 3.6905,
+      "step": 23860
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.8649181383000122e-05,
+      "loss": 3.9271,
+      "step": 23865
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.8608954503399173e-05,
+      "loss": 2.276,
+      "step": 23870
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.8568727623798225e-05,
+      "loss": 1.4453,
+      "step": 23875
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.8528500744197273e-05,
+      "loss": 1.511,
+      "step": 23880
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.8488273864596324e-05,
+      "loss": 1.0376,
+      "step": 23885
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.8448046984995373e-05,
+      "loss": 4.748,
+      "step": 23890
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.8407820105394424e-05,
+      "loss": 3.6019,
+      "step": 23895
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.8367593225793475e-05,
+      "loss": 4.3068,
+      "step": 23900
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.8327366346192527e-05,
+      "loss": 3.8338,
+      "step": 23905
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.828713946659158e-05,
+      "loss": 1.3604,
+      "step": 23910
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.8246912586990626e-05,
+      "loss": 2.0121,
+      "step": 23915
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.8206685707389678e-05,
+      "loss": 3.177,
+      "step": 23920
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.816645882778873e-05,
+      "loss": 2.4406,
+      "step": 23925
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.812623194818778e-05,
+      "loss": 2.1305,
+      "step": 23930
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.8086005068586832e-05,
+      "loss": 0.9836,
+      "step": 23935
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.804577818898588e-05,
+      "loss": 4.1375,
+      "step": 23940
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.8005551309384932e-05,
+      "loss": 3.9156,
+      "step": 23945
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.7965324429783983e-05,
+      "loss": 3.7585,
+      "step": 23950
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.7925097550183035e-05,
+      "loss": 3.27,
+      "step": 23955
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.7884870670582086e-05,
+      "loss": 4.2422,
+      "step": 23960
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.7844643790981134e-05,
+      "loss": 3.0293,
+      "step": 23965
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.7804416911380182e-05,
+      "loss": 3.3175,
+      "step": 23970
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.7764190031779234e-05,
+      "loss": 2.4429,
+      "step": 23975
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.7723963152178285e-05,
+      "loss": 1.251,
+      "step": 23980
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.7683736272577337e-05,
+      "loss": 2.908,
+      "step": 23985
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.764350939297639e-05,
+      "loss": 4.0328,
+      "step": 23990
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.7603282513375436e-05,
+      "loss": 3.9133,
+      "step": 23995
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.7563055633774488e-05,
+      "loss": 4.8342,
+      "step": 24000
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.752282875417354e-05,
+      "loss": 3.8279,
+      "step": 24005
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.748260187457259e-05,
+      "loss": 1.9251,
+      "step": 24010
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.7442374994971642e-05,
+      "loss": 2.9262,
+      "step": 24015
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.7402148115370694e-05,
+      "loss": 2.1363,
+      "step": 24020
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.7361921235769742e-05,
+      "loss": 1.7946,
+      "step": 24025
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.7321694356168793e-05,
+      "loss": 1.5292,
+      "step": 24030
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.7281467476567845e-05,
+      "loss": 4.4959,
+      "step": 24035
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.7241240596966893e-05,
+      "loss": 4.6133,
+      "step": 24040
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.7201013717365944e-05,
+      "loss": 4.9404,
+      "step": 24045
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.7160786837764996e-05,
+      "loss": 3.764,
+      "step": 24050
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.7120559958164044e-05,
+      "loss": 4.8266,
+      "step": 24055
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.7080333078563095e-05,
+      "loss": 3.7552,
+      "step": 24060
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.7040106198962147e-05,
+      "loss": 2.8064,
+      "step": 24065
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6999879319361198e-05,
+      "loss": 3.0139,
+      "step": 24070
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.695965243976025e-05,
+      "loss": 2.4623,
+      "step": 24075
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6919425560159298e-05,
+      "loss": 3.624,
+      "step": 24080
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.687919868055835e-05,
+      "loss": 1.5064,
+      "step": 24085
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.68389718009574e-05,
+      "loss": 4.409,
+      "step": 24090
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6798744921356452e-05,
+      "loss": 5.5877,
+      "step": 24095
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6758518041755504e-05,
+      "loss": 3.7055,
+      "step": 24100
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6718291162154555e-05,
+      "loss": 2.6408,
+      "step": 24105
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6678064282553603e-05,
+      "loss": 4.0404,
+      "step": 24110
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6637837402952655e-05,
+      "loss": 2.8409,
+      "step": 24115
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6597610523351703e-05,
+      "loss": 2.9476,
+      "step": 24120
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6557383643750754e-05,
+      "loss": 0.8873,
+      "step": 24125
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6517156764149806e-05,
+      "loss": 2.8307,
+      "step": 24130
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6476929884548857e-05,
+      "loss": 2.2032,
+      "step": 24135
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.6436703004947905e-05,
+      "loss": 4.2484,
+      "step": 24140
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.6396476125346957e-05,
+      "loss": 3.6621,
+      "step": 24145
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.6356249245746008e-05,
+      "loss": 4.5232,
+      "step": 24150
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.631602236614506e-05,
+      "loss": 3.4658,
+      "step": 24155
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.627579548654411e-05,
+      "loss": 3.6425,
+      "step": 24160
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.623556860694316e-05,
+      "loss": 2.3619,
+      "step": 24165
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.619534172734221e-05,
+      "loss": 2.8156,
+      "step": 24170
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.6155114847741262e-05,
+      "loss": 0.9569,
+      "step": 24175
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.6114887968140314e-05,
+      "loss": 1.9015,
+      "step": 24180
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.6074661088539365e-05,
+      "loss": 1.9707,
+      "step": 24185
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.6034434208938413e-05,
+      "loss": 4.2664,
+      "step": 24190
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.599420732933746e-05,
+      "loss": 4.7297,
+      "step": 24195
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.5953980449736513e-05,
+      "loss": 3.9335,
+      "step": 24200
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.5913753570135564e-05,
+      "loss": 3.4213,
+      "step": 24205
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.5873526690534616e-05,
+      "loss": 3.0822,
+      "step": 24210
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.5833299810933667e-05,
+      "loss": 2.5482,
+      "step": 24215
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.579307293133272e-05,
+      "loss": 3.5275,
+      "step": 24220
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.5752846051731767e-05,
+      "loss": 2.0448,
+      "step": 24225
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.5712619172130818e-05,
+      "loss": 1.7843,
+      "step": 24230
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.567239229252987e-05,
+      "loss": 1.5835,
+      "step": 24235
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.563216541292892e-05,
+      "loss": 4.832,
+      "step": 24240
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.5591938533327973e-05,
+      "loss": 3.8062,
+      "step": 24245
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.555171165372702e-05,
+      "loss": 3.6148,
+      "step": 24250
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.5511484774126072e-05,
+      "loss": 4.2596,
+      "step": 24255
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.5471257894525124e-05,
+      "loss": 3.2226,
+      "step": 24260
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.5431031014924175e-05,
+      "loss": 2.3329,
+      "step": 24265
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.5390804135323223e-05,
+      "loss": 3.3771,
+      "step": 24270
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.5350577255722275e-05,
+      "loss": 2.6768,
+      "step": 24275
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.5310350376121323e-05,
+      "loss": 3.1228,
+      "step": 24280
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.5270123496520374e-05,
+      "loss": 1.9548,
+      "step": 24285
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.5229896616919426e-05,
+      "loss": 4.4299,
+      "step": 24290
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.5189669737318477e-05,
+      "loss": 4.6846,
+      "step": 24295
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.5149442857717527e-05,
+      "loss": 4.5248,
+      "step": 24300
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.5109215978116578e-05,
+      "loss": 3.1845,
+      "step": 24305
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.506898909851563e-05,
+      "loss": 2.2644,
+      "step": 24310
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.502876221891468e-05,
+      "loss": 1.6771,
+      "step": 24315
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4988535339313731e-05,
+      "loss": 2.1373,
+      "step": 24320
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.494830845971278e-05,
+      "loss": 1.9354,
+      "step": 24325
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4908081580111832e-05,
+      "loss": 1.3599,
+      "step": 24330
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4867854700510884e-05,
+      "loss": 1.6992,
+      "step": 24335
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4827627820909933e-05,
+      "loss": 4.1914,
+      "step": 24340
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4787400941308982e-05,
+      "loss": 5.009,
+      "step": 24345
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4747174061708033e-05,
+      "loss": 3.6693,
+      "step": 24350
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4706947182107084e-05,
+      "loss": 3.537,
+      "step": 24355
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4666720302506134e-05,
+      "loss": 2.5959,
+      "step": 24360
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4626493422905186e-05,
+      "loss": 2.8873,
+      "step": 24365
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4586266543304235e-05,
+      "loss": 2.0024,
+      "step": 24370
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4546039663703287e-05,
+      "loss": 3.3242,
+      "step": 24375
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4505812784102338e-05,
+      "loss": 1.5681,
+      "step": 24380
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4465585904501388e-05,
+      "loss": 4.3365,
+      "step": 24385
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.442535902490044e-05,
+      "loss": 4.5049,
+      "step": 24390
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4385132145299491e-05,
+      "loss": 3.7834,
+      "step": 24395
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4344905265698541e-05,
+      "loss": 3.6518,
+      "step": 24400
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4304678386097592e-05,
+      "loss": 3.5904,
+      "step": 24405
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4264451506496642e-05,
+      "loss": 2.5824,
+      "step": 24410
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4224224626895694e-05,
+      "loss": 3.6668,
+      "step": 24415
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4183997747294745e-05,
+      "loss": 2.6437,
+      "step": 24420
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4143770867693793e-05,
+      "loss": 1.704,
+      "step": 24425
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4103543988092843e-05,
+      "loss": 2.013,
+      "step": 24430
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4063317108491894e-05,
+      "loss": 0.8377,
+      "step": 24435
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4023090228890944e-05,
+      "loss": 4.3678,
+      "step": 24440
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.3982863349289996e-05,
+      "loss": 4.1889,
+      "step": 24445
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.3942636469689047e-05,
+      "loss": 4.3562,
+      "step": 24450
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.3902409590088097e-05,
+      "loss": 4.6929,
+      "step": 24455
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.3862182710487148e-05,
+      "loss": 3.6932,
+      "step": 24460
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.38219558308862e-05,
+      "loss": 2.9506,
+      "step": 24465
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.378172895128525e-05,
+      "loss": 1.9533,
+      "step": 24470
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.3741502071684301e-05,
+      "loss": 1.3351,
+      "step": 24475
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.370127519208335e-05,
+      "loss": 2.0779,
+      "step": 24480
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.3661048312482402e-05,
+      "loss": 1.5529,
+      "step": 24485
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.3620821432881454e-05,
+      "loss": 4.1068,
+      "step": 24490
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.3580594553280504e-05,
+      "loss": 4.8277,
+      "step": 24495
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.3540367673679552e-05,
+      "loss": 3.857,
+      "step": 24500
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.3500140794078603e-05,
+      "loss": 4.1428,
+      "step": 24505
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.3459913914477655e-05,
+      "loss": 3.019,
+      "step": 24510
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.3419687034876704e-05,
+      "loss": 2.5749,
+      "step": 24515
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.3379460155275756e-05,
+      "loss": 3.4932,
+      "step": 24520
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.3339233275674806e-05,
+      "loss": 2.0479,
+      "step": 24525
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.3299006396073857e-05,
+      "loss": 3.4595,
+      "step": 24530
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.3258779516472908e-05,
+      "loss": 2.5689,
+      "step": 24535
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.3218552636871958e-05,
+      "loss": 3.9021,
+      "step": 24540
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.317832575727101e-05,
+      "loss": 4.1131,
+      "step": 24545
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.313809887767006e-05,
+      "loss": 3.3045,
+      "step": 24550
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.3097871998069111e-05,
+      "loss": 3.8982,
+      "step": 24555
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.3057645118468162e-05,
+      "loss": 2.1605,
+      "step": 24560
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.3017418238867212e-05,
+      "loss": 3.1492,
+      "step": 24565
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.2977191359266264e-05,
+      "loss": 1.427,
+      "step": 24570
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.2936964479665312e-05,
+      "loss": 2.1544,
+      "step": 24575
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.2896737600064363e-05,
+      "loss": 3.2604,
+      "step": 24580
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.2856510720463413e-05,
+      "loss": 1.2599,
+      "step": 24585
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.2816283840862464e-05,
+      "loss": 4.0691,
+      "step": 24590
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.2776056961261514e-05,
+      "loss": 4.2889,
+      "step": 24595
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.2735830081660566e-05,
+      "loss": 3.4133,
+      "step": 24600
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.2695603202059617e-05,
+      "loss": 4.1035,
+      "step": 24605
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.2655376322458667e-05,
+      "loss": 3.4185,
+      "step": 24610
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.2615149442857718e-05,
+      "loss": 2.4912,
+      "step": 24615
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.257492256325677e-05,
+      "loss": 2.678,
+      "step": 24620
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.253469568365582e-05,
+      "loss": 1.9975,
+      "step": 24625
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.2494468804054871e-05,
+      "loss": 0.6405,
+      "step": 24630
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.2454241924453921e-05,
+      "loss": 1.8826,
+      "step": 24635
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.241401504485297e-05,
+      "loss": 4.4268,
+      "step": 24640
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.2373788165252022e-05,
+      "loss": 3.8896,
+      "step": 24645
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.2333561285651072e-05,
+      "loss": 4.8965,
+      "step": 24650
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.2293334406050123e-05,
+      "loss": 3.1094,
+      "step": 24655
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.2253107526449175e-05,
+      "loss": 2.9067,
+      "step": 24660
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.2212880646848225e-05,
+      "loss": 2.9209,
+      "step": 24665
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.2172653767247274e-05,
+      "loss": 2.7946,
+      "step": 24670
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.2132426887646326e-05,
+      "loss": 1.3476,
+      "step": 24675
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.2092200008045376e-05,
+      "loss": 1.7138,
+      "step": 24680
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.2051973128444427e-05,
+      "loss": 2.9379,
+      "step": 24685
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.2011746248843479e-05,
+      "loss": 3.7029,
+      "step": 24690
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1971519369242528e-05,
+      "loss": 4.3479,
+      "step": 24695
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.193129248964158e-05,
+      "loss": 3.7926,
+      "step": 24700
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.189106561004063e-05,
+      "loss": 3.3841,
+      "step": 24705
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.185083873043968e-05,
+      "loss": 4.1531,
+      "step": 24710
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.181061185083873e-05,
+      "loss": 2.6995,
+      "step": 24715
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1770384971237782e-05,
+      "loss": 1.0784,
+      "step": 24720
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1730158091636832e-05,
+      "loss": 2.5422,
+      "step": 24725
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1689931212035884e-05,
+      "loss": 2.6564,
+      "step": 24730
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1649704332434933e-05,
+      "loss": 3.302,
+      "step": 24735
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1609477452833985e-05,
+      "loss": 3.8146,
+      "step": 24740
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1569250573233035e-05,
+      "loss": 3.9229,
+      "step": 24745
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1529023693632084e-05,
+      "loss": 4.4492,
+      "step": 24750
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1488796814031136e-05,
+      "loss": 3.6742,
+      "step": 24755
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1448569934430187e-05,
+      "loss": 2.4106,
+      "step": 24760
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1408343054829237e-05,
+      "loss": 2.7309,
+      "step": 24765
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1368116175228288e-05,
+      "loss": 1.6355,
+      "step": 24770
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.132788929562734e-05,
+      "loss": 2.7919,
+      "step": 24775
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.128766241602639e-05,
+      "loss": 4.5257,
+      "step": 24780
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.124743553642544e-05,
+      "loss": 3.9398,
+      "step": 24785
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1207208656824491e-05,
+      "loss": 4.6109,
+      "step": 24790
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.116698177722354e-05,
+      "loss": 3.3293,
+      "step": 24795
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1126754897622592e-05,
+      "loss": 3.921,
+      "step": 24800
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1086528018021642e-05,
+      "loss": 4.2807,
+      "step": 24805
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1046301138420693e-05,
+      "loss": 3.1406,
+      "step": 24810
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1006074258819745e-05,
+      "loss": 3.4969,
+      "step": 24815
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.0965847379218795e-05,
+      "loss": 2.3031,
+      "step": 24820
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.0925620499617844e-05,
+      "loss": 3.3691,
+      "step": 24825
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.0885393620016896e-05,
+      "loss": 2.8064,
+      "step": 24830
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.0845166740415946e-05,
+      "loss": 1.1896,
+      "step": 24835
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.0804939860814997e-05,
+      "loss": 4.5449,
+      "step": 24840
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.0764712981214049e-05,
+      "loss": 3.3299,
+      "step": 24845
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.0724486101613098e-05,
+      "loss": 4.4686,
+      "step": 24850
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.068425922201215e-05,
+      "loss": 3.9811,
+      "step": 24855
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.06440323424112e-05,
+      "loss": 3.383,
+      "step": 24860
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.060380546281025e-05,
+      "loss": 2.5898,
+      "step": 24865
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.0563578583209301e-05,
+      "loss": 2.4245,
+      "step": 24870
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.052335170360835e-05,
+      "loss": 2.3957,
+      "step": 24875
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.0483124824007402e-05,
+      "loss": 1.7501,
+      "step": 24880
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.0442897944406454e-05,
+      "loss": 1.7959,
+      "step": 24885
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.0402671064805503e-05,
+      "loss": 4.0732,
+      "step": 24890
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.0362444185204555e-05,
+      "loss": 4.116,
+      "step": 24895
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.0322217305603605e-05,
+      "loss": 4.5895,
+      "step": 24900
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.0281990426002654e-05,
+      "loss": 3.9937,
+      "step": 24905
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.0241763546401706e-05,
+      "loss": 4.1619,
+      "step": 24910
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.0201536666800757e-05,
+      "loss": 4.0855,
+      "step": 24915
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.0161309787199807e-05,
+      "loss": 2.9171,
+      "step": 24920
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.0121082907598859e-05,
+      "loss": 1.7636,
+      "step": 24925
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.008085602799791e-05,
+      "loss": 2.8449,
+      "step": 24930
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.004062914839696e-05,
+      "loss": 2.8344,
+      "step": 24935
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.000040226879601e-05,
+      "loss": 3.7783,
+      "step": 24940
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.960175389195061e-06,
+      "loss": 4.4023,
+      "step": 24945
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.91994850959411e-06,
+      "loss": 3.5332,
+      "step": 24950
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.879721629993162e-06,
+      "loss": 2.6086,
+      "step": 24955
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.839494750392212e-06,
+      "loss": 3.8582,
+      "step": 24960
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.799267870791264e-06,
+      "loss": 2.062,
+      "step": 24965
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.759040991190315e-06,
+      "loss": 3.416,
+      "step": 24970
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.718814111589363e-06,
+      "loss": 2.2982,
+      "step": 24975
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.678587231988415e-06,
+      "loss": 0.8549,
+      "step": 24980
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.638360352387466e-06,
+      "loss": 2.1826,
+      "step": 24985
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.598133472786516e-06,
+      "loss": 5.0311,
+      "step": 24990
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.557906593185567e-06,
+      "loss": 4.0307,
+      "step": 24995
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.517679713584619e-06,
+      "loss": 3.3335,
+      "step": 25000
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.477452833983668e-06,
+      "loss": 2.483,
+      "step": 25005
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.43722595438272e-06,
+      "loss": 3.2506,
+      "step": 25010
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.39699907478177e-06,
+      "loss": 3.9112,
+      "step": 25015
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.35677219518082e-06,
+      "loss": 2.9277,
+      "step": 25020
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.316545315579871e-06,
+      "loss": 1.3599,
+      "step": 25025
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.27631843597892e-06,
+      "loss": 0.6077,
+      "step": 25030
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.236091556377972e-06,
+      "loss": 1.0871,
+      "step": 25035
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.195864676777024e-06,
+      "loss": 3.9141,
+      "step": 25040
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.155637797176073e-06,
+      "loss": 4.5723,
+      "step": 25045
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.115410917575123e-06,
+      "loss": 3.6136,
+      "step": 25050
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.075184037974175e-06,
+      "loss": 3.3955,
+      "step": 25055
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.034957158373224e-06,
+      "loss": 3.2014,
+      "step": 25060
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.994730278772276e-06,
+      "loss": 2.9881,
+      "step": 25065
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.954503399171327e-06,
+      "loss": 3.3361,
+      "step": 25070
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.914276519570377e-06,
+      "loss": 2.0125,
+      "step": 25075
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.874049639969429e-06,
+      "loss": 2.2374,
+      "step": 25080
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.833822760368478e-06,
+      "loss": 2.4366,
+      "step": 25085
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.793595880767528e-06,
+      "loss": 3.9693,
+      "step": 25090
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.75336900116658e-06,
+      "loss": 4.5941,
+      "step": 25095
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.713142121565631e-06,
+      "loss": 4.5533,
+      "step": 25100
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.672915241964681e-06,
+      "loss": 4.7197,
+      "step": 25105
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.632688362363732e-06,
+      "loss": 3.6422,
+      "step": 25110
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.592461482762782e-06,
+      "loss": 2.0502,
+      "step": 25115
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.552234603161834e-06,
+      "loss": 3.1903,
+      "step": 25120
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.512007723560885e-06,
+      "loss": 1.453,
+      "step": 25125
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.471780843959933e-06,
+      "loss": 2.6905,
+      "step": 25130
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.431553964358985e-06,
+      "loss": 1.6192,
+      "step": 25135
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.391327084758036e-06,
+      "loss": 3.7004,
+      "step": 25140
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.351100205157086e-06,
+      "loss": 3.7824,
+      "step": 25145
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.310873325556137e-06,
+      "loss": 5.5062,
+      "step": 25150
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.270646445955189e-06,
+      "loss": 3.8611,
+      "step": 25155
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.230419566354239e-06,
+      "loss": 3.7684,
+      "step": 25160
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.190192686753288e-06,
+      "loss": 3.4285,
+      "step": 25165
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.14996580715234e-06,
+      "loss": 3.1098,
+      "step": 25170
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.10973892755139e-06,
+      "loss": 3.1606,
+      "step": 25175
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.069512047950441e-06,
+      "loss": 1.8826,
+      "step": 25180
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.02928516834949e-06,
+      "loss": 2.1713,
+      "step": 25185
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.989058288748542e-06,
+      "loss": 4.7242,
+      "step": 25190
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.948831409147594e-06,
+      "loss": 4.0975,
+      "step": 25195
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.908604529546644e-06,
+      "loss": 3.651,
+      "step": 25200
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.868377649945693e-06,
+      "loss": 4.6125,
+      "step": 25205
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.828150770344745e-06,
+      "loss": 3.1507,
+      "step": 25210
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.787923890743795e-06,
+      "loss": 2.712,
+      "step": 25215
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.747697011142846e-06,
+      "loss": 3.0889,
+      "step": 25220
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.707470131541897e-06,
+      "loss": 1.9682,
+      "step": 25225
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.667243251940947e-06,
+      "loss": 3.9404,
+      "step": 25230
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.627016372339999e-06,
+      "loss": 2.6924,
+      "step": 25235
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.586789492739049e-06,
+      "loss": 4.6404,
+      "step": 25240
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.546562613138099e-06,
+      "loss": 4.0658,
+      "step": 25245
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.50633573353715e-06,
+      "loss": 3.5027,
+      "step": 25250
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.4661088539362e-06,
+      "loss": 4.2753,
+      "step": 25255
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.425881974335251e-06,
+      "loss": 2.2604,
+      "step": 25260
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.385655094734302e-06,
+      "loss": 2.5162,
+      "step": 25265
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.345428215133353e-06,
+      "loss": 3.1524,
+      "step": 25270
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.305201335532404e-06,
+      "loss": 2.0495,
+      "step": 25275
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.2649744559314534e-06,
+      "loss": 2.8922,
+      "step": 25280
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.224747576330504e-06,
+      "loss": 0.9049,
+      "step": 25285
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.184520696729555e-06,
+      "loss": 3.7818,
+      "step": 25290
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.144293817128605e-06,
+      "loss": 3.5922,
+      "step": 25295
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.104066937527657e-06,
+      "loss": 3.708,
+      "step": 25300
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.063840057926707e-06,
+      "loss": 3.2295,
+      "step": 25305
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.023613178325758e-06,
+      "loss": 3.4177,
+      "step": 25310
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.983386298724809e-06,
+      "loss": 2.8741,
+      "step": 25315
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.943159419123858e-06,
+      "loss": 2.726,
+      "step": 25320
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.902932539522909e-06,
+      "loss": 3.9523,
+      "step": 25325
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.8707510358421505e-06,
+      "loss": 3.2671,
+      "step": 25330
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.830524156241201e-06,
+      "loss": 3.2882,
+      "step": 25335
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.790297276640252e-06,
+      "loss": 3.732,
+      "step": 25340
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.7500703970393015e-06,
+      "loss": 5.0133,
+      "step": 25345
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.709843517438352e-06,
+      "loss": 2.8558,
+      "step": 25350
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.669616637837403e-06,
+      "loss": 3.9777,
+      "step": 25355
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.629389758236454e-06,
+      "loss": 2.9651,
+      "step": 25360
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.589162878635505e-06,
+      "loss": 3.3267,
+      "step": 25365
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.5489359990345555e-06,
+      "loss": 2.6909,
+      "step": 25370
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.508709119433606e-06,
+      "loss": 1.3714,
+      "step": 25375
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.468482239832656e-06,
+      "loss": 3.6478,
+      "step": 25380
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.4282553602317065e-06,
+      "loss": 0.7816,
+      "step": 25385
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.388028480630757e-06,
+      "loss": 3.7924,
+      "step": 25390
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.347801601029809e-06,
+      "loss": 3.7242,
+      "step": 25395
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.307574721428859e-06,
+      "loss": 5.2104,
+      "step": 25400
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.26734784182791e-06,
+      "loss": 3.7268,
+      "step": 25405
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.2271209622269604e-06,
+      "loss": 3.1541,
+      "step": 25410
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.186894082626011e-06,
+      "loss": 4.2297,
+      "step": 25415
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.146667203025062e-06,
+      "loss": 2.2711,
+      "step": 25420
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.106440323424112e-06,
+      "loss": 3.1184,
+      "step": 25425
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.066213443823163e-06,
+      "loss": 2.7837,
+      "step": 25430
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.0259865642222135e-06,
+      "loss": 1.4746,
+      "step": 25435
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.985759684621264e-06,
+      "loss": 5.2686,
+      "step": 25440
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.945532805020315e-06,
+      "loss": 4.7791,
+      "step": 25445
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.905305925419365e-06,
+      "loss": 3.8869,
+      "step": 25450
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.865079045818416e-06,
+      "loss": 3.2348,
+      "step": 25455
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.824852166217467e-06,
+      "loss": 4.1375,
+      "step": 25460
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.784625286616517e-06,
+      "loss": 2.7724,
+      "step": 25465
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.744398407015568e-06,
+      "loss": 2.3747,
+      "step": 25470
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7041715274146185e-06,
+      "loss": 1.9912,
+      "step": 25475
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.66394464781367e-06,
+      "loss": 2.2611,
+      "step": 25480
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.62371776821272e-06,
+      "loss": 2.0734,
+      "step": 25485
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.58349088861177e-06,
+      "loss": 4.0169,
+      "step": 25490
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.543264009010821e-06,
+      "loss": 4.0657,
+      "step": 25495
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.5030371294098725e-06,
+      "loss": 3.7256,
+      "step": 25500
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.462810249808922e-06,
+      "loss": 3.7052,
+      "step": 25505
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.422583370207973e-06,
+      "loss": 3.3589,
+      "step": 25510
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.382356490607024e-06,
+      "loss": 3.0988,
+      "step": 25515
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.342129611006075e-06,
+      "loss": 1.339,
+      "step": 25520
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.301902731405125e-06,
+      "loss": 2.7499,
+      "step": 25525
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.261675851804175e-06,
+      "loss": 2.7686,
+      "step": 25530
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.221448972203227e-06,
+      "loss": 3.5382,
+      "step": 25535
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.181222092602277e-06,
+      "loss": 3.9172,
+      "step": 25540
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.140995213001327e-06,
+      "loss": 4.0725,
+      "step": 25545
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.100768333400379e-06,
+      "loss": 4.2646,
+      "step": 25550
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.060541453799429e-06,
+      "loss": 3.8309,
+      "step": 25555
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.02031457419848e-06,
+      "loss": 3.0889,
+      "step": 25560
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.9800876945975305e-06,
+      "loss": 3.4834,
+      "step": 25565
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.939860814996581e-06,
+      "loss": 3.4956,
+      "step": 25570
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.899633935395632e-06,
+      "loss": 1.8908,
+      "step": 25575
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.8594070557946815e-06,
+      "loss": 4.3148,
+      "step": 25580
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.819180176193733e-06,
+      "loss": 3.6078,
+      "step": 25585
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.778953296592784e-06,
+      "loss": 3.7635,
+      "step": 25590
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.738726416991834e-06,
+      "loss": 3.8357,
+      "step": 25595
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.698499537390885e-06,
+      "loss": 3.5507,
+      "step": 25600
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.6582726577899355e-06,
+      "loss": 2.5274,
+      "step": 25605
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.618045778188986e-06,
+      "loss": 2.7862,
+      "step": 25610
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.577818898588037e-06,
+      "loss": 2.16,
+      "step": 25615
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.537592018987087e-06,
+      "loss": 3.5172,
+      "step": 25620
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.497365139386138e-06,
+      "loss": 1.672,
+      "step": 25625
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.4571382597851886e-06,
+      "loss": 0.9123,
+      "step": 25630
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.416911380184239e-06,
+      "loss": 1.4239,
+      "step": 25635
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.37668450058329e-06,
+      "loss": 4.6506,
+      "step": 25640
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.3364576209823404e-06,
+      "loss": 3.7811,
+      "step": 25645
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.296230741381391e-06,
+      "loss": 3.5947,
+      "step": 25650
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.2560038617804425e-06,
+      "loss": 3.8759,
+      "step": 25655
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.215776982179492e-06,
+      "loss": 3.2007,
+      "step": 25660
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.175550102578543e-06,
+      "loss": 2.9636,
+      "step": 25665
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.135323222977594e-06,
+      "loss": 2.5895,
+      "step": 25670
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.095096343376644e-06,
+      "loss": 2.5369,
+      "step": 25675
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.054869463775695e-06,
+      "loss": 1.8998,
+      "step": 25680
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.014642584174745e-06,
+      "loss": 3.4223,
+      "step": 25685
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.974415704573797e-06,
+      "loss": 4.2734,
+      "step": 25690
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.934188824972847e-06,
+      "loss": 3.8895,
+      "step": 25695
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.893961945371897e-06,
+      "loss": 4.1982,
+      "step": 25700
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.853735065770949e-06,
+      "loss": 3.1064,
+      "step": 25705
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.8135081861699993e-06,
+      "loss": 3.512,
+      "step": 25710
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.7732813065690495e-06,
+      "loss": 2.4828,
+      "step": 25715
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.7330544269681e-06,
+      "loss": 3.1687,
+      "step": 25720
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.692827547367151e-06,
+      "loss": 3.2518,
+      "step": 25725
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.652600667766202e-06,
+      "loss": 2.2701,
+      "step": 25730
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.612373788165252e-06,
+      "loss": 2.698,
+      "step": 25735
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.5721469085643026e-06,
+      "loss": 4.4486,
+      "step": 25740
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.5319200289633537e-06,
+      "loss": 4.2291,
+      "step": 25745
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.4916931493624043e-06,
+      "loss": 3.1553,
+      "step": 25750
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.4514662697614545e-06,
+      "loss": 4.0023,
+      "step": 25755
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.4112393901605055e-06,
+      "loss": 5.1721,
+      "step": 25760
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.371012510559556e-06,
+      "loss": 2.4705,
+      "step": 25765
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.330785630958607e-06,
+      "loss": 2.4637,
+      "step": 25770
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.290558751357657e-06,
+      "loss": 2.728,
+      "step": 25775
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.250331871756708e-06,
+      "loss": 2.834,
+      "step": 25780
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.2101049921557586e-06,
+      "loss": 1.8569,
+      "step": 25785
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.169878112554809e-06,
+      "loss": 4.0803,
+      "step": 25790
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.12965123295386e-06,
+      "loss": 3.2705,
+      "step": 25795
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.0894243533529105e-06,
+      "loss": 3.3291,
+      "step": 25800
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.049197473751961e-06,
+      "loss": 5.2221,
+      "step": 25805
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.0089705941510117e-06,
+      "loss": 3.5207,
+      "step": 25810
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.9687437145500624e-06,
+      "loss": 3.5691,
+      "step": 25815
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.9285168349491134e-06,
+      "loss": 1.4946,
+      "step": 25820
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.8882899553481636e-06,
+      "loss": 2.3518,
+      "step": 25825
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.8480630757472147e-06,
+      "loss": 2.2248,
+      "step": 25830
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.807836196146265e-06,
+      "loss": 2.2885,
+      "step": 25835
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.767609316545316e-06,
+      "loss": 3.764,
+      "step": 25840
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.7273824369443665e-06,
+      "loss": 4.1377,
+      "step": 25845
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.687155557343417e-06,
+      "loss": 4.3615,
+      "step": 25850
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.6469286777424678e-06,
+      "loss": 3.9139,
+      "step": 25855
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.6067017981415184e-06,
+      "loss": 3.9283,
+      "step": 25860
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.566474918540569e-06,
+      "loss": 2.8451,
+      "step": 25865
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.5262480389396196e-06,
+      "loss": 2.6068,
+      "step": 25870
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.4860211593386702e-06,
+      "loss": 3.0293,
+      "step": 25875
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.445794279737721e-06,
+      "loss": 1.631,
+      "step": 25880
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.4055674001367715e-06,
+      "loss": 1.9995,
+      "step": 25885
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.365340520535822e-06,
+      "loss": 4.6109,
+      "step": 25890
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.3251136409348727e-06,
+      "loss": 4.3992,
+      "step": 25895
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.2848867613339233e-06,
+      "loss": 3.3955,
+      "step": 25900
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.244659881732974e-06,
+      "loss": 2.3407,
+      "step": 25905
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.2044330021320246e-06,
+      "loss": 2.9357,
+      "step": 25910
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.1642061225310756e-06,
+      "loss": 2.0345,
+      "step": 25915
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.123979242930126e-06,
+      "loss": 2.8767,
+      "step": 25920
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.083752363329177e-06,
+      "loss": 2.4327,
+      "step": 25925
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.043525483728227e-06,
+      "loss": 1.5194,
+      "step": 25930
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.003298604127278e-06,
+      "loss": 2.0557,
+      "step": 25935
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9630717245263287e-06,
+      "loss": 4.5334,
+      "step": 25940
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9228448449253793e-06,
+      "loss": 3.535,
+      "step": 25945
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.8826179653244297e-06,
+      "loss": 3.9789,
+      "step": 25950
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.8423910857234806e-06,
+      "loss": 4.2756,
+      "step": 25955
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.8021642061225312e-06,
+      "loss": 3.7443,
+      "step": 25960
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.761937326521582e-06,
+      "loss": 2.4496,
+      "step": 25965
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.7217104469206324e-06,
+      "loss": 2.5245,
+      "step": 25970
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6814835673196833e-06,
+      "loss": 2.2765,
+      "step": 25975
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6412566877187337e-06,
+      "loss": 2.3552,
+      "step": 25980
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6010298081177843e-06,
+      "loss": 3.3843,
+      "step": 25985
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.5608029285168351e-06,
+      "loss": 4.0086,
+      "step": 25990
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.5205760489158858e-06,
+      "loss": 4.1575,
+      "step": 25995
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.4803491693149364e-06,
+      "loss": 4.2672,
+      "step": 26000
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.440122289713987e-06,
+      "loss": 3.7455,
+      "step": 26005
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3998954101130376e-06,
+      "loss": 3.8271,
+      "step": 26010
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3596685305120882e-06,
+      "loss": 3.1133,
+      "step": 26015
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3194416509111389e-06,
+      "loss": 3.2707,
+      "step": 26020
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.2792147713101897e-06,
+      "loss": 1.7346,
+      "step": 26025
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.23898789170924e-06,
+      "loss": 1.4858,
+      "step": 26030
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.1987610121082907e-06,
+      "loss": 2.2901,
+      "step": 26035
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.1585341325073413e-06,
+      "loss": 4.3547,
+      "step": 26040
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.118307252906392e-06,
+      "loss": 4.3891,
+      "step": 26045
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0780803733054428e-06,
+      "loss": 3.8504,
+      "step": 26050
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0378534937044934e-06,
+      "loss": 3.1188,
+      "step": 26055
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.97626614103544e-07,
+      "loss": 3.6335,
+      "step": 26060
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.573997345025946e-07,
+      "loss": 1.9387,
+      "step": 26065
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.171728549016453e-07,
+      "loss": 3.2563,
+      "step": 26070
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 8.76945975300696e-07,
+      "loss": 2.0166,
+      "step": 26075
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 8.367190956997466e-07,
+      "loss": 3.0024,
+      "step": 26080
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 7.964922160987972e-07,
+      "loss": 3.026,
+      "step": 26085
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 7.56265336497848e-07,
+      "loss": 4.6111,
+      "step": 26090
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 7.160384568968986e-07,
+      "loss": 4.4326,
+      "step": 26095
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.758115772959492e-07,
+      "loss": 4.4527,
+      "step": 26100
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.355846976949998e-07,
+      "loss": 2.902,
+      "step": 26105
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.953578180940504e-07,
+      "loss": 3.3514,
+      "step": 26110
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.551309384931011e-07,
+      "loss": 3.167,
+      "step": 26115
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.149040588921518e-07,
+      "loss": 2.4512,
+      "step": 26120
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.746771792912024e-07,
+      "loss": 1.9114,
+      "step": 26125
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.3445029969025303e-07,
+      "loss": 2.0456,
+      "step": 26130
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.942234200893037e-07,
+      "loss": 1.6755,
+      "step": 26135
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.5399654048835437e-07,
+      "loss": 5.232,
+      "step": 26140
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.13769660887405e-07,
+      "loss": 3.2018,
+      "step": 26145
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.735427812864556e-07,
+      "loss": 2.7205,
+      "step": 26150
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.3331590168550626e-07,
+      "loss": 3.5699,
+      "step": 26155
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.930890220845569e-07,
+      "loss": 3.0736,
+      "step": 26160
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.5286214248360755e-07,
+      "loss": 1.3811,
+      "step": 26165
+    },
+    {
+      "epoch": 2.0,
+      "eval_exact_match": 19.45205479452055,
+      "eval_f1": 23.848435348153973,
+      "eval_loss": 3.1337833404541016,
+      "eval_runtime": 131.2034,
+      "eval_samples_per_second": 11.128,
+      "eval_steps_per_second": 11.128,
+      "step": 26168
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 26168,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "save_steps": 500,
+  "total_flos": 8847303388182072.0,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}