diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -1,8 +1,8 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 96.3302752293578,
-  "global_step": 10500,
+  "epoch": 120.0,
+  "global_step": 13080,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
@@ -63218,18 +63218,15548 @@
       "step": 10500
     },
     {
-      "epoch": 96.33,
-      "step": 10500,
-      "total_flos": 1.2873670788396168e+20,
-      "train_loss": 0.0,
-      "train_runtime": 113.9955,
-      "train_samples_per_second": 152.55,
-      "train_steps_per_second": 4.781
+      "epoch": 96.34,
+      "learning_rate": 9.873853211009175e-05,
+      "loss": 0.1608,
+      "step": 10501
+    },
+    {
+      "epoch": 96.35,
+      "learning_rate": 9.870030581039755e-05,
+      "loss": 0.1755,
+      "step": 10502
+    },
+    {
+      "epoch": 96.36,
+      "learning_rate": 9.866207951070337e-05,
+      "loss": 0.0978,
+      "step": 10503
+    },
+    {
+      "epoch": 96.37,
+      "learning_rate": 9.862385321100918e-05,
+      "loss": 0.1589,
+      "step": 10504
+    },
+    {
+      "epoch": 96.38,
+      "learning_rate": 9.8585626911315e-05,
+      "loss": 0.1309,
+      "step": 10505
+    },
+    {
+      "epoch": 96.39,
+      "learning_rate": 9.854740061162078e-05,
+      "loss": 0.1038,
+      "step": 10506
+    },
+    {
+      "epoch": 96.39,
+      "learning_rate": 9.85091743119266e-05,
+      "loss": 0.0773,
+      "step": 10507
+    },
+    {
+      "epoch": 96.4,
+      "learning_rate": 9.847094801223242e-05,
+      "loss": 0.0701,
+      "step": 10508
+    },
+    {
+      "epoch": 96.41,
+      "learning_rate": 9.843272171253822e-05,
+      "loss": 0.0693,
+      "step": 10509
+    },
+    {
+      "epoch": 96.42,
+      "learning_rate": 9.839449541284404e-05,
+      "loss": 0.1357,
+      "step": 10510
+    },
+    {
+      "epoch": 96.43,
+      "learning_rate": 9.835626911314985e-05,
+      "loss": 0.0942,
+      "step": 10511
+    },
+    {
+      "epoch": 96.44,
+      "learning_rate": 9.831804281345566e-05,
+      "loss": 0.1105,
+      "step": 10512
+    },
+    {
+      "epoch": 96.45,
+      "learning_rate": 9.827981651376147e-05,
+      "loss": 0.1312,
+      "step": 10513
+    },
+    {
+      "epoch": 96.46,
+      "learning_rate": 9.824159021406729e-05,
+      "loss": 0.1684,
+      "step": 10514
+    },
+    {
+      "epoch": 96.47,
+      "learning_rate": 9.820336391437309e-05,
+      "loss": 0.0683,
+      "step": 10515
+    },
+    {
+      "epoch": 96.48,
+      "learning_rate": 9.816513761467891e-05,
+      "loss": 0.1148,
+      "step": 10516
+    },
+    {
+      "epoch": 96.49,
+      "learning_rate": 9.812691131498471e-05,
+      "loss": 0.0958,
+      "step": 10517
+    },
+    {
+      "epoch": 96.5,
+      "learning_rate": 9.808868501529052e-05,
+      "loss": 0.1225,
+      "step": 10518
+    },
+    {
+      "epoch": 96.5,
+      "learning_rate": 9.805045871559632e-05,
+      "loss": 0.1302,
+      "step": 10519
+    },
+    {
+      "epoch": 96.51,
+      "learning_rate": 9.801223241590214e-05,
+      "loss": 0.1702,
+      "step": 10520
+    },
+    {
+      "epoch": 96.52,
+      "learning_rate": 9.797400611620794e-05,
+      "loss": 0.1444,
+      "step": 10521
+    },
+    {
+      "epoch": 96.53,
+      "learning_rate": 9.793577981651376e-05,
+      "loss": 0.1642,
+      "step": 10522
+    },
+    {
+      "epoch": 96.54,
+      "learning_rate": 9.789755351681958e-05,
+      "loss": 0.1328,
+      "step": 10523
+    },
+    {
+      "epoch": 96.55,
+      "learning_rate": 9.785932721712538e-05,
+      "loss": 0.1216,
+      "step": 10524
+    },
+    {
+      "epoch": 96.56,
+      "learning_rate": 9.78211009174312e-05,
+      "loss": 0.1151,
+      "step": 10525
+    },
+    {
+      "epoch": 96.57,
+      "learning_rate": 9.7782874617737e-05,
+      "loss": 0.0657,
+      "step": 10526
+    },
+    {
+      "epoch": 96.58,
+      "learning_rate": 9.774464831804282e-05,
+      "loss": 0.0773,
+      "step": 10527
+    },
+    {
+      "epoch": 96.59,
+      "learning_rate": 9.770642201834863e-05,
+      "loss": 0.1217,
+      "step": 10528
+    },
+    {
+      "epoch": 96.6,
+      "learning_rate": 9.766819571865445e-05,
+      "loss": 0.0877,
+      "step": 10529
+    },
+    {
+      "epoch": 96.61,
+      "learning_rate": 9.762996941896024e-05,
+      "loss": 0.1168,
+      "step": 10530
+    },
+    {
+      "epoch": 96.61,
+      "learning_rate": 9.759174311926605e-05,
+      "loss": 0.0754,
+      "step": 10531
+    },
+    {
+      "epoch": 96.62,
+      "learning_rate": 9.755351681957186e-05,
+      "loss": 0.0723,
+      "step": 10532
+    },
+    {
+      "epoch": 96.63,
+      "learning_rate": 9.751529051987768e-05,
+      "loss": 0.1496,
+      "step": 10533
+    },
+    {
+      "epoch": 96.64,
+      "learning_rate": 9.747706422018348e-05,
+      "loss": 0.1072,
+      "step": 10534
+    },
+    {
+      "epoch": 96.65,
+      "learning_rate": 9.74388379204893e-05,
+      "loss": 0.141,
+      "step": 10535
+    },
+    {
+      "epoch": 96.66,
+      "learning_rate": 9.740061162079512e-05,
+      "loss": 0.0756,
+      "step": 10536
+    },
+    {
+      "epoch": 96.67,
+      "learning_rate": 9.736238532110092e-05,
+      "loss": 0.1152,
+      "step": 10537
+    },
+    {
+      "epoch": 96.68,
+      "learning_rate": 9.732415902140674e-05,
+      "loss": 0.1491,
+      "step": 10538
+    },
+    {
+      "epoch": 96.69,
+      "learning_rate": 9.728593272171254e-05,
+      "loss": 0.0857,
+      "step": 10539
+    },
+    {
+      "epoch": 96.7,
+      "learning_rate": 9.724770642201836e-05,
+      "loss": 0.0887,
+      "step": 10540
+    },
+    {
+      "epoch": 96.71,
+      "learning_rate": 9.720948012232417e-05,
+      "loss": 0.0449,
+      "step": 10541
+    },
+    {
+      "epoch": 96.72,
+      "learning_rate": 9.717125382262997e-05,
+      "loss": 0.0635,
+      "step": 10542
+    },
+    {
+      "epoch": 96.72,
+      "learning_rate": 9.713302752293577e-05,
+      "loss": 0.0883,
+      "step": 10543
+    },
+    {
+      "epoch": 96.73,
+      "learning_rate": 9.709480122324159e-05,
+      "loss": 0.1412,
+      "step": 10544
+    },
+    {
+      "epoch": 96.74,
+      "learning_rate": 9.70565749235474e-05,
+      "loss": 0.143,
+      "step": 10545
+    },
+    {
+      "epoch": 96.75,
+      "learning_rate": 9.701834862385321e-05,
+      "loss": 0.1546,
+      "step": 10546
+    },
+    {
+      "epoch": 96.76,
+      "learning_rate": 9.698012232415902e-05,
+      "loss": 0.1197,
+      "step": 10547
+    },
+    {
+      "epoch": 96.77,
+      "learning_rate": 9.694189602446484e-05,
+      "loss": 0.1359,
+      "step": 10548
+    },
+    {
+      "epoch": 96.78,
+      "learning_rate": 9.690366972477064e-05,
+      "loss": 0.1124,
+      "step": 10549
+    },
+    {
+      "epoch": 96.79,
+      "learning_rate": 9.686544342507646e-05,
+      "loss": 0.1273,
+      "step": 10550
+    },
+    {
+      "epoch": 96.8,
+      "learning_rate": 9.682721712538228e-05,
+      "loss": 0.1145,
+      "step": 10551
+    },
+    {
+      "epoch": 96.81,
+      "learning_rate": 9.678899082568808e-05,
+      "loss": 0.0947,
+      "step": 10552
+    },
+    {
+      "epoch": 96.82,
+      "learning_rate": 9.67507645259939e-05,
+      "loss": 0.0917,
+      "step": 10553
+    },
+    {
+      "epoch": 96.83,
+      "learning_rate": 9.671253822629969e-05,
+      "loss": 0.1067,
+      "step": 10554
+    },
+    {
+      "epoch": 96.83,
+      "learning_rate": 9.66743119266055e-05,
+      "loss": 0.0778,
+      "step": 10555
+    },
+    {
+      "epoch": 96.84,
+      "learning_rate": 9.663608562691131e-05,
+      "loss": 0.0993,
+      "step": 10556
+    },
+    {
+      "epoch": 96.85,
+      "learning_rate": 9.659785932721713e-05,
+      "loss": 0.0875,
+      "step": 10557
+    },
+    {
+      "epoch": 96.86,
+      "learning_rate": 9.655963302752293e-05,
+      "loss": 0.0887,
+      "step": 10558
+    },
+    {
+      "epoch": 96.87,
+      "learning_rate": 9.652140672782875e-05,
+      "loss": 0.0695,
+      "step": 10559
+    },
+    {
+      "epoch": 96.88,
+      "learning_rate": 9.648318042813455e-05,
+      "loss": 0.1077,
+      "step": 10560
+    },
+    {
+      "epoch": 96.89,
+      "learning_rate": 9.644495412844037e-05,
+      "loss": 0.1115,
+      "step": 10561
+    },
+    {
+      "epoch": 96.9,
+      "learning_rate": 9.640672782874618e-05,
+      "loss": 0.135,
+      "step": 10562
+    },
+    {
+      "epoch": 96.91,
+      "learning_rate": 9.6368501529052e-05,
+      "loss": 0.0785,
+      "step": 10563
+    },
+    {
+      "epoch": 96.92,
+      "learning_rate": 9.63302752293578e-05,
+      "loss": 0.142,
+      "step": 10564
+    },
+    {
+      "epoch": 96.93,
+      "learning_rate": 9.629204892966362e-05,
+      "loss": 0.1055,
+      "step": 10565
+    },
+    {
+      "epoch": 96.94,
+      "learning_rate": 9.625382262996942e-05,
+      "loss": 0.1395,
+      "step": 10566
+    },
+    {
+      "epoch": 96.94,
+      "learning_rate": 9.621559633027523e-05,
+      "loss": 0.1324,
+      "step": 10567
+    },
+    {
+      "epoch": 96.95,
+      "learning_rate": 9.617737003058104e-05,
+      "loss": 0.0849,
+      "step": 10568
+    },
+    {
+      "epoch": 96.96,
+      "learning_rate": 9.613914373088685e-05,
+      "loss": 0.1442,
+      "step": 10569
+    },
+    {
+      "epoch": 96.97,
+      "learning_rate": 9.610091743119267e-05,
+      "loss": 0.1004,
+      "step": 10570
+    },
+    {
+      "epoch": 96.98,
+      "learning_rate": 9.606269113149847e-05,
+      "loss": 0.1033,
+      "step": 10571
+    },
+    {
+      "epoch": 96.99,
+      "learning_rate": 9.602446483180429e-05,
+      "loss": 0.1199,
+      "step": 10572
+    },
+    {
+      "epoch": 97.0,
+      "learning_rate": 9.598623853211009e-05,
+      "loss": 0.0849,
+      "step": 10573
+    },
+    {
+      "epoch": 97.01,
+      "learning_rate": 9.594801223241591e-05,
+      "loss": 0.1637,
+      "step": 10574
+    },
+    {
+      "epoch": 97.02,
+      "learning_rate": 9.590978593272171e-05,
+      "loss": 0.135,
+      "step": 10575
+    },
+    {
+      "epoch": 97.03,
+      "learning_rate": 9.587155963302753e-05,
+      "loss": 0.1242,
+      "step": 10576
+    },
+    {
+      "epoch": 97.04,
+      "learning_rate": 9.583333333333334e-05,
+      "loss": 0.1277,
+      "step": 10577
+    },
+    {
+      "epoch": 97.05,
+      "learning_rate": 9.579510703363914e-05,
+      "loss": 0.1904,
+      "step": 10578
+    },
+    {
+      "epoch": 97.06,
+      "learning_rate": 9.575688073394494e-05,
+      "loss": 0.1065,
+      "step": 10579
+    },
+    {
+      "epoch": 97.06,
+      "learning_rate": 9.571865443425076e-05,
+      "loss": 0.0857,
+      "step": 10580
+    },
+    {
+      "epoch": 97.07,
+      "learning_rate": 9.568042813455658e-05,
+      "loss": 0.1157,
+      "step": 10581
+    },
+    {
+      "epoch": 97.08,
+      "learning_rate": 9.564220183486238e-05,
+      "loss": 0.1409,
+      "step": 10582
+    },
+    {
+      "epoch": 97.09,
+      "learning_rate": 9.56039755351682e-05,
+      "loss": 0.1194,
+      "step": 10583
+    },
+    {
+      "epoch": 97.1,
+      "learning_rate": 9.556574923547401e-05,
+      "loss": 0.1039,
+      "step": 10584
+    },
+    {
+      "epoch": 97.11,
+      "learning_rate": 9.552752293577982e-05,
+      "loss": 0.1274,
+      "step": 10585
+    },
+    {
+      "epoch": 97.12,
+      "learning_rate": 9.548929663608563e-05,
+      "loss": 0.1331,
+      "step": 10586
+    },
+    {
+      "epoch": 97.13,
+      "learning_rate": 9.545107033639145e-05,
+      "loss": 0.0853,
+      "step": 10587
+    },
+    {
+      "epoch": 97.14,
+      "learning_rate": 9.541284403669725e-05,
+      "loss": 0.0938,
+      "step": 10588
+    },
+    {
+      "epoch": 97.15,
+      "learning_rate": 9.537461773700307e-05,
+      "loss": 0.0918,
+      "step": 10589
+    },
+    {
+      "epoch": 97.16,
+      "learning_rate": 9.533639143730886e-05,
+      "loss": 0.1407,
+      "step": 10590
+    },
+    {
+      "epoch": 97.17,
+      "learning_rate": 9.529816513761468e-05,
+      "loss": 0.1013,
+      "step": 10591
+    },
+    {
+      "epoch": 97.17,
+      "learning_rate": 9.525993883792048e-05,
+      "loss": 0.0936,
+      "step": 10592
+    },
+    {
+      "epoch": 97.18,
+      "learning_rate": 9.52217125382263e-05,
+      "loss": 0.1746,
+      "step": 10593
+    },
+    {
+      "epoch": 97.19,
+      "learning_rate": 9.51834862385321e-05,
+      "loss": 0.1102,
+      "step": 10594
+    },
+    {
+      "epoch": 97.2,
+      "learning_rate": 9.514525993883792e-05,
+      "loss": 0.1698,
+      "step": 10595
+    },
+    {
+      "epoch": 97.21,
+      "learning_rate": 9.510703363914374e-05,
+      "loss": 0.0952,
+      "step": 10596
+    },
+    {
+      "epoch": 97.22,
+      "learning_rate": 9.506880733944954e-05,
+      "loss": 0.1334,
+      "step": 10597
+    },
+    {
+      "epoch": 97.23,
+      "learning_rate": 9.503058103975536e-05,
+      "loss": 0.1066,
+      "step": 10598
+    },
+    {
+      "epoch": 97.24,
+      "learning_rate": 9.499235474006117e-05,
+      "loss": 0.1187,
+      "step": 10599
+    },
+    {
+      "epoch": 97.25,
+      "learning_rate": 9.495412844036698e-05,
+      "loss": 0.0963,
+      "step": 10600
+    },
+    {
+      "epoch": 97.26,
+      "learning_rate": 9.491590214067279e-05,
+      "loss": 0.1839,
+      "step": 10601
+    },
+    {
+      "epoch": 97.27,
+      "learning_rate": 9.487767584097859e-05,
+      "loss": 0.1484,
+      "step": 10602
+    },
+    {
+      "epoch": 97.28,
+      "learning_rate": 9.48394495412844e-05,
+      "loss": 0.1716,
+      "step": 10603
+    },
+    {
+      "epoch": 97.28,
+      "learning_rate": 9.480122324159021e-05,
+      "loss": 0.1419,
+      "step": 10604
+    },
+    {
+      "epoch": 97.29,
+      "learning_rate": 9.476299694189602e-05,
+      "loss": 0.1404,
+      "step": 10605
+    },
+    {
+      "epoch": 97.3,
+      "learning_rate": 9.472477064220184e-05,
+      "loss": 0.1667,
+      "step": 10606
+    },
+    {
+      "epoch": 97.31,
+      "learning_rate": 9.468654434250764e-05,
+      "loss": 0.1162,
+      "step": 10607
+    },
+    {
+      "epoch": 97.32,
+      "learning_rate": 9.464831804281346e-05,
+      "loss": 0.0987,
+      "step": 10608
+    },
+    {
+      "epoch": 97.33,
+      "learning_rate": 9.461009174311926e-05,
+      "loss": 0.1306,
+      "step": 10609
+    },
+    {
+      "epoch": 97.34,
+      "learning_rate": 9.457186544342508e-05,
+      "loss": 0.0856,
+      "step": 10610
+    },
+    {
+      "epoch": 97.35,
+      "learning_rate": 9.45336391437309e-05,
+      "loss": 0.1073,
+      "step": 10611
+    },
+    {
+      "epoch": 97.36,
+      "learning_rate": 9.44954128440367e-05,
+      "loss": 0.0922,
+      "step": 10612
+    },
+    {
+      "epoch": 97.37,
+      "learning_rate": 9.445718654434252e-05,
+      "loss": 0.1296,
+      "step": 10613
+    },
+    {
+      "epoch": 97.38,
+      "learning_rate": 9.441896024464831e-05,
+      "loss": 0.1814,
+      "step": 10614
+    },
+    {
+      "epoch": 97.39,
+      "learning_rate": 9.438073394495413e-05,
+      "loss": 0.1006,
+      "step": 10615
+    },
+    {
+      "epoch": 97.39,
+      "learning_rate": 9.434250764525993e-05,
+      "loss": 0.1703,
+      "step": 10616
+    },
+    {
+      "epoch": 97.4,
+      "learning_rate": 9.430428134556575e-05,
+      "loss": 0.1023,
+      "step": 10617
+    },
+    {
+      "epoch": 97.41,
+      "learning_rate": 9.426605504587156e-05,
+      "loss": 0.1091,
+      "step": 10618
+    },
+    {
+      "epoch": 97.42,
+      "learning_rate": 9.422782874617737e-05,
+      "loss": 0.1287,
+      "step": 10619
+    },
+    {
+      "epoch": 97.43,
+      "learning_rate": 9.418960244648318e-05,
+      "loss": 0.1212,
+      "step": 10620
+    },
+    {
+      "epoch": 97.44,
+      "learning_rate": 9.4151376146789e-05,
+      "loss": 0.215,
+      "step": 10621
+    },
+    {
+      "epoch": 97.45,
+      "learning_rate": 9.41131498470948e-05,
+      "loss": 0.1602,
+      "step": 10622
+    },
+    {
+      "epoch": 97.46,
+      "learning_rate": 9.407492354740062e-05,
+      "loss": 0.0687,
+      "step": 10623
+    },
+    {
+      "epoch": 97.47,
+      "learning_rate": 9.403669724770644e-05,
+      "loss": 0.1355,
+      "step": 10624
+    },
+    {
+      "epoch": 97.48,
+      "learning_rate": 9.399847094801224e-05,
+      "loss": 0.0946,
+      "step": 10625
+    },
+    {
+      "epoch": 97.49,
+      "learning_rate": 9.396024464831804e-05,
+      "loss": 0.0603,
+      "step": 10626
+    },
+    {
+      "epoch": 97.5,
+      "learning_rate": 9.392201834862385e-05,
+      "loss": 0.0516,
+      "step": 10627
+    },
+    {
+      "epoch": 97.5,
+      "learning_rate": 9.388379204892967e-05,
+      "loss": 0.1725,
+      "step": 10628
+    },
+    {
+      "epoch": 97.51,
+      "learning_rate": 9.384556574923547e-05,
+      "loss": 0.1566,
+      "step": 10629
+    },
+    {
+      "epoch": 97.52,
+      "learning_rate": 9.380733944954129e-05,
+      "loss": 0.1736,
+      "step": 10630
+    },
+    {
+      "epoch": 97.53,
+      "learning_rate": 9.376911314984709e-05,
+      "loss": 0.1354,
+      "step": 10631
+    },
+    {
+      "epoch": 97.54,
+      "learning_rate": 9.373088685015291e-05,
+      "loss": 0.1036,
+      "step": 10632
+    },
+    {
+      "epoch": 97.55,
+      "learning_rate": 9.369266055045872e-05,
+      "loss": 0.0817,
+      "step": 10633
+    },
+    {
+      "epoch": 97.56,
+      "learning_rate": 9.365443425076453e-05,
+      "loss": 0.1002,
+      "step": 10634
+    },
+    {
+      "epoch": 97.57,
+      "learning_rate": 9.361620795107034e-05,
+      "loss": 0.1238,
+      "step": 10635
+    },
+    {
+      "epoch": 97.58,
+      "learning_rate": 9.357798165137616e-05,
+      "loss": 0.0902,
+      "step": 10636
+    },
+    {
+      "epoch": 97.59,
+      "learning_rate": 9.353975535168196e-05,
+      "loss": 0.09,
+      "step": 10637
+    },
+    {
+      "epoch": 97.6,
+      "learning_rate": 9.350152905198776e-05,
+      "loss": 0.1204,
+      "step": 10638
+    },
+    {
+      "epoch": 97.61,
+      "learning_rate": 9.346330275229357e-05,
+      "loss": 0.0981,
+      "step": 10639
+    },
+    {
+      "epoch": 97.61,
+      "learning_rate": 9.342507645259939e-05,
+      "loss": 0.1004,
+      "step": 10640
+    },
+    {
+      "epoch": 97.62,
+      "learning_rate": 9.33868501529052e-05,
+      "loss": 0.1542,
+      "step": 10641
+    },
+    {
+      "epoch": 97.63,
+      "learning_rate": 9.334862385321101e-05,
+      "loss": 0.1231,
+      "step": 10642
+    },
+    {
+      "epoch": 97.64,
+      "learning_rate": 9.331039755351683e-05,
+      "loss": 0.1078,
+      "step": 10643
+    },
+    {
+      "epoch": 97.65,
+      "learning_rate": 9.327217125382263e-05,
+      "loss": 0.1091,
+      "step": 10644
+    },
+    {
+      "epoch": 97.66,
+      "learning_rate": 9.323394495412845e-05,
+      "loss": 0.1045,
+      "step": 10645
+    },
+    {
+      "epoch": 97.67,
+      "learning_rate": 9.319571865443425e-05,
+      "loss": 0.0999,
+      "step": 10646
+    },
+    {
+      "epoch": 97.68,
+      "learning_rate": 9.315749235474007e-05,
+      "loss": 0.1106,
+      "step": 10647
+    },
+    {
+      "epoch": 97.69,
+      "learning_rate": 9.311926605504587e-05,
+      "loss": 0.1104,
+      "step": 10648
+    },
+    {
+      "epoch": 97.7,
+      "learning_rate": 9.308103975535169e-05,
+      "loss": 0.0963,
+      "step": 10649
+    },
+    {
+      "epoch": 97.71,
+      "learning_rate": 9.304281345565748e-05,
+      "loss": 0.1252,
+      "step": 10650
+    },
+    {
+      "epoch": 97.72,
+      "learning_rate": 9.30045871559633e-05,
+      "loss": 0.0703,
+      "step": 10651
+    },
+    {
+      "epoch": 97.72,
+      "learning_rate": 9.29663608562691e-05,
+      "loss": 0.1013,
+      "step": 10652
+    },
+    {
+      "epoch": 97.73,
+      "learning_rate": 9.292813455657492e-05,
+      "loss": 0.1037,
+      "step": 10653
+    },
+    {
+      "epoch": 97.74,
+      "learning_rate": 9.288990825688074e-05,
+      "loss": 0.0295,
+      "step": 10654
+    },
+    {
+      "epoch": 97.75,
+      "learning_rate": 9.285168195718655e-05,
+      "loss": 0.1799,
+      "step": 10655
+    },
+    {
+      "epoch": 97.76,
+      "learning_rate": 9.281345565749236e-05,
+      "loss": 0.1424,
+      "step": 10656
+    },
+    {
+      "epoch": 97.77,
+      "learning_rate": 9.277522935779817e-05,
+      "loss": 0.1223,
+      "step": 10657
+    },
+    {
+      "epoch": 97.78,
+      "learning_rate": 9.273700305810398e-05,
+      "loss": 0.1,
+      "step": 10658
+    },
+    {
+      "epoch": 97.79,
+      "learning_rate": 9.269877675840979e-05,
+      "loss": 0.129,
+      "step": 10659
+    },
+    {
+      "epoch": 97.8,
+      "learning_rate": 9.266055045871561e-05,
+      "loss": 0.1149,
+      "step": 10660
+    },
+    {
+      "epoch": 97.81,
+      "learning_rate": 9.262232415902141e-05,
+      "loss": 0.1512,
+      "step": 10661
+    },
+    {
+      "epoch": 97.82,
+      "learning_rate": 9.258409785932722e-05,
+      "loss": 0.1736,
+      "step": 10662
+    },
+    {
+      "epoch": 97.83,
+      "learning_rate": 9.254587155963302e-05,
+      "loss": 0.092,
+      "step": 10663
+    },
+    {
+      "epoch": 97.83,
+      "learning_rate": 9.250764525993884e-05,
+      "loss": 0.1595,
+      "step": 10664
+    },
+    {
+      "epoch": 97.84,
+      "learning_rate": 9.246941896024464e-05,
+      "loss": 0.1275,
+      "step": 10665
+    },
+    {
+      "epoch": 97.85,
+      "learning_rate": 9.243119266055046e-05,
+      "loss": 0.1404,
+      "step": 10666
+    },
+    {
+      "epoch": 97.86,
+      "learning_rate": 9.239296636085626e-05,
+      "loss": 0.1071,
+      "step": 10667
+    },
+    {
+      "epoch": 97.87,
+      "learning_rate": 9.235474006116208e-05,
+      "loss": 0.15,
+      "step": 10668
+    },
+    {
+      "epoch": 97.88,
+      "learning_rate": 9.23165137614679e-05,
+      "loss": 0.1968,
+      "step": 10669
+    },
+    {
+      "epoch": 97.89,
+      "learning_rate": 9.22782874617737e-05,
+      "loss": 0.1037,
+      "step": 10670
+    },
+    {
+      "epoch": 97.9,
+      "learning_rate": 9.224006116207952e-05,
+      "loss": 0.1442,
+      "step": 10671
+    },
+    {
+      "epoch": 97.91,
+      "learning_rate": 9.220183486238533e-05,
+      "loss": 0.145,
+      "step": 10672
+    },
+    {
+      "epoch": 97.92,
+      "learning_rate": 9.216360856269114e-05,
+      "loss": 0.0877,
+      "step": 10673
+    },
+    {
+      "epoch": 97.93,
+      "learning_rate": 9.212538226299693e-05,
+      "loss": 0.0991,
+      "step": 10674
+    },
+    {
+      "epoch": 97.94,
+      "learning_rate": 9.208715596330275e-05,
+      "loss": 0.1057,
+      "step": 10675
+    },
+    {
+      "epoch": 97.94,
+      "learning_rate": 9.204892966360856e-05,
+      "loss": 0.082,
+      "step": 10676
+    },
+    {
+      "epoch": 97.95,
+      "learning_rate": 9.201070336391437e-05,
+      "loss": 0.1038,
+      "step": 10677
+    },
+    {
+      "epoch": 97.96,
+      "learning_rate": 9.197247706422018e-05,
+      "loss": 0.1058,
+      "step": 10678
+    },
+    {
+      "epoch": 97.97,
+      "learning_rate": 9.1934250764526e-05,
+      "loss": 0.1413,
+      "step": 10679
+    },
+    {
+      "epoch": 97.98,
+      "learning_rate": 9.18960244648318e-05,
+      "loss": 0.1176,
+      "step": 10680
+    },
+    {
+      "epoch": 97.99,
+      "learning_rate": 9.185779816513762e-05,
+      "loss": 0.0353,
+      "step": 10681
+    },
+    {
+      "epoch": 98.0,
+      "learning_rate": 9.181957186544342e-05,
+      "loss": 0.0801,
+      "step": 10682
+    },
+    {
+      "epoch": 98.01,
+      "learning_rate": 9.178134556574924e-05,
+      "loss": 0.2246,
+      "step": 10683
+    },
+    {
+      "epoch": 98.02,
+      "learning_rate": 9.174311926605506e-05,
+      "loss": 0.1257,
+      "step": 10684
+    },
+    {
+      "epoch": 98.03,
+      "learning_rate": 9.170489296636086e-05,
+      "loss": 0.1353,
+      "step": 10685
+    },
+    {
+      "epoch": 98.04,
+      "learning_rate": 9.166666666666667e-05,
+      "loss": 0.1157,
+      "step": 10686
+    },
+    {
+      "epoch": 98.05,
+      "learning_rate": 9.162844036697247e-05,
+      "loss": 0.171,
+      "step": 10687
+    },
+    {
+      "epoch": 98.06,
+      "learning_rate": 9.159021406727829e-05,
+      "loss": 0.1292,
+      "step": 10688
+    },
+    {
+      "epoch": 98.06,
+      "learning_rate": 9.15519877675841e-05,
+      "loss": 0.1498,
+      "step": 10689
+    },
+    {
+      "epoch": 98.07,
+      "learning_rate": 9.151376146788991e-05,
+      "loss": 0.1411,
+      "step": 10690
+    },
+    {
+      "epoch": 98.08,
+      "learning_rate": 9.147553516819572e-05,
+      "loss": 0.1448,
+      "step": 10691
+    },
+    {
+      "epoch": 98.09,
+      "learning_rate": 9.143730886850153e-05,
+      "loss": 0.1244,
+      "step": 10692
+    },
+    {
+      "epoch": 98.1,
+      "learning_rate": 9.139908256880734e-05,
+      "loss": 0.1452,
+      "step": 10693
+    },
+    {
+      "epoch": 98.11,
+      "learning_rate": 9.136085626911316e-05,
+      "loss": 0.1093,
+      "step": 10694
+    },
+    {
+      "epoch": 98.12,
+      "learning_rate": 9.132262996941896e-05,
+      "loss": 0.1058,
+      "step": 10695
+    },
+    {
+      "epoch": 98.13,
+      "learning_rate": 9.128440366972478e-05,
+      "loss": 0.1084,
+      "step": 10696
+    },
+    {
+      "epoch": 98.14,
+      "learning_rate": 9.124617737003058e-05,
+      "loss": 0.1302,
+      "step": 10697
+    },
+    {
+      "epoch": 98.15,
+      "learning_rate": 9.120795107033639e-05,
+      "loss": 0.1432,
+      "step": 10698
+    },
+    {
+      "epoch": 98.16,
+      "learning_rate": 9.11697247706422e-05,
+      "loss": 0.0703,
+      "step": 10699
+    },
+    {
+      "epoch": 98.17,
+      "learning_rate": 9.113149847094801e-05,
+      "loss": 0.1469,
+      "step": 10700
+    },
+    {
+      "epoch": 98.17,
+      "learning_rate": 9.109327217125383e-05,
+      "loss": 0.0913,
+      "step": 10701
+    },
+    {
+      "epoch": 98.18,
+      "learning_rate": 9.105504587155963e-05,
+      "loss": 0.1069,
+      "step": 10702
+    },
+    {
+      "epoch": 98.19,
+      "learning_rate": 9.101681957186545e-05,
+      "loss": 0.1676,
+      "step": 10703
+    },
+    {
+      "epoch": 98.2,
+      "learning_rate": 9.097859327217125e-05,
+      "loss": 0.0309,
+      "step": 10704
+    },
+    {
+      "epoch": 98.21,
+      "learning_rate": 9.094036697247707e-05,
+      "loss": 0.0768,
+      "step": 10705
+    },
+    {
+      "epoch": 98.22,
+      "learning_rate": 9.090214067278288e-05,
+      "loss": 0.1327,
+      "step": 10706
+    },
+    {
+      "epoch": 98.23,
+      "learning_rate": 9.08639143730887e-05,
+      "loss": 0.0996,
+      "step": 10707
+    },
+    {
+      "epoch": 98.24,
+      "learning_rate": 9.08256880733945e-05,
+      "loss": 0.1094,
+      "step": 10708
+    },
+    {
+      "epoch": 98.25,
+      "learning_rate": 9.078746177370032e-05,
+      "loss": 0.0903,
+      "step": 10709
+    },
+    {
+      "epoch": 98.26,
+      "learning_rate": 9.07492354740061e-05,
+      "loss": 0.1773,
+      "step": 10710
+    },
+    {
+      "epoch": 98.27,
+      "learning_rate": 9.071100917431192e-05,
+      "loss": 0.1687,
+      "step": 10711
+    },
+    {
+      "epoch": 98.28,
+      "learning_rate": 9.067278287461773e-05,
+      "loss": 0.1288,
+      "step": 10712
+    },
+    {
+      "epoch": 98.28,
+      "learning_rate": 9.063455657492355e-05,
+      "loss": 0.1103,
+      "step": 10713
+    },
+    {
+      "epoch": 98.29,
+      "learning_rate": 9.059633027522936e-05,
+      "loss": 0.1706,
+      "step": 10714
+    },
+    {
+      "epoch": 98.3,
+      "learning_rate": 9.055810397553517e-05,
+      "loss": 0.1381,
+      "step": 10715
+    },
+    {
+      "epoch": 98.31,
+      "learning_rate": 9.051987767584099e-05,
+      "loss": 0.108,
+      "step": 10716
+    },
+    {
+      "epoch": 98.32,
+      "learning_rate": 9.048165137614679e-05,
+      "loss": 0.1368,
+      "step": 10717
+    },
+    {
+      "epoch": 98.33,
+      "learning_rate": 9.044342507645261e-05,
+      "loss": 0.1698,
+      "step": 10718
+    },
+    {
+      "epoch": 98.34,
+      "learning_rate": 9.040519877675841e-05,
+      "loss": 0.1133,
+      "step": 10719
+    },
+    {
+      "epoch": 98.35,
+      "learning_rate": 9.036697247706423e-05,
+      "loss": 0.145,
+      "step": 10720
+    },
+    {
+      "epoch": 98.36,
+      "learning_rate": 9.032874617737003e-05,
+      "loss": 0.1637,
+      "step": 10721
+    },
+    {
+      "epoch": 98.37,
+      "learning_rate": 9.029051987767584e-05,
+      "loss": 0.1439,
+      "step": 10722
+    },
+    {
+      "epoch": 98.38,
+      "learning_rate": 9.025229357798164e-05,
+      "loss": 0.1301,
+      "step": 10723
+    },
+    {
+      "epoch": 98.39,
+      "learning_rate": 9.021406727828746e-05,
+      "loss": 0.1295,
+      "step": 10724
+    },
+    {
+      "epoch": 98.39,
+      "learning_rate": 9.017584097859327e-05,
+      "loss": 0.1513,
+      "step": 10725
+    },
+    {
+      "epoch": 98.4,
+      "learning_rate": 9.013761467889908e-05,
+      "loss": 0.1343,
+      "step": 10726
+    },
+    {
+      "epoch": 98.41,
+      "learning_rate": 9.009938837920489e-05,
+      "loss": 0.1275,
+      "step": 10727
+    },
+    {
+      "epoch": 98.42,
+      "learning_rate": 9.00611620795107e-05,
+      "loss": 0.105,
+      "step": 10728
+    },
+    {
+      "epoch": 98.43,
+      "learning_rate": 9.002293577981652e-05,
+      "loss": 0.1102,
+      "step": 10729
+    },
+    {
+      "epoch": 98.44,
+      "learning_rate": 8.998470948012233e-05,
+      "loss": 0.1167,
+      "step": 10730
+    },
+    {
+      "epoch": 98.45,
+      "learning_rate": 8.994648318042815e-05,
+      "loss": 0.0779,
+      "step": 10731
+    },
+    {
+      "epoch": 98.46,
+      "learning_rate": 8.990825688073395e-05,
+      "loss": 0.0744,
+      "step": 10732
+    },
+    {
+      "epoch": 98.47,
+      "learning_rate": 8.987003058103977e-05,
+      "loss": 0.1967,
+      "step": 10733
+    },
+    {
+      "epoch": 98.48,
+      "learning_rate": 8.983180428134556e-05,
+      "loss": 0.1088,
+      "step": 10734
+    },
+    {
+      "epoch": 98.49,
+      "learning_rate": 8.979357798165138e-05,
+      "loss": 0.0881,
+      "step": 10735
+    },
+    {
+      "epoch": 98.5,
+      "learning_rate": 8.975535168195718e-05,
+      "loss": 0.0291,
+      "step": 10736
+    },
+    {
+      "epoch": 98.5,
+      "learning_rate": 8.9717125382263e-05,
+      "loss": 0.179,
+      "step": 10737
+    },
+    {
+      "epoch": 98.51,
+      "learning_rate": 8.96788990825688e-05,
+      "loss": 0.1708,
+      "step": 10738
+    },
+    {
+      "epoch": 98.52,
+      "learning_rate": 8.964067278287462e-05,
+      "loss": 0.1168,
+      "step": 10739
+    },
+    {
+      "epoch": 98.53,
+      "learning_rate": 8.960244648318042e-05,
+      "loss": 0.1364,
+      "step": 10740
+    },
+    {
+      "epoch": 98.54,
+      "learning_rate": 8.956422018348624e-05,
+      "loss": 0.1683,
+      "step": 10741
+    },
+    {
+      "epoch": 98.55,
+      "learning_rate": 8.952599388379206e-05,
+      "loss": 0.1355,
+      "step": 10742
+    },
+    {
+      "epoch": 98.56,
+      "learning_rate": 8.948776758409786e-05,
+      "loss": 0.1637,
+      "step": 10743
+    },
+    {
+      "epoch": 98.57,
+      "learning_rate": 8.944954128440368e-05,
+      "loss": 0.1265,
+      "step": 10744
+    },
+    {
+      "epoch": 98.58,
+      "learning_rate": 8.941131498470949e-05,
+      "loss": 0.1094,
+      "step": 10745
+    },
+    {
+      "epoch": 98.59,
+      "learning_rate": 8.937308868501529e-05,
+      "loss": 0.1575,
+      "step": 10746
+    },
+    {
+      "epoch": 98.6,
+      "learning_rate": 8.93348623853211e-05,
+      "loss": 0.1514,
+      "step": 10747
+    },
+    {
+      "epoch": 98.61,
+      "learning_rate": 8.929663608562691e-05,
+      "loss": 0.1838,
+      "step": 10748
+    },
+    {
+      "epoch": 98.61,
+      "learning_rate": 8.925840978593272e-05,
+      "loss": 0.0822,
+      "step": 10749
+    },
+    {
+      "epoch": 98.62,
+      "learning_rate": 8.922018348623854e-05,
+      "loss": 0.1029,
+      "step": 10750
+    },
+    {
+      "epoch": 98.63,
+      "learning_rate": 8.918195718654434e-05,
+      "loss": 0.1407,
+      "step": 10751
+    },
+    {
+      "epoch": 98.64,
+      "learning_rate": 8.914373088685016e-05,
+      "loss": 0.0543,
+      "step": 10752
+    },
+    {
+      "epoch": 98.65,
+      "learning_rate": 8.910550458715596e-05,
+      "loss": 0.0513,
+      "step": 10753
+    },
+    {
+      "epoch": 98.66,
+      "learning_rate": 8.906727828746178e-05,
+      "loss": 0.1082,
+      "step": 10754
+    },
+    {
+      "epoch": 98.67,
+      "learning_rate": 8.902905198776758e-05,
+      "loss": 0.1045,
+      "step": 10755
+    },
+    {
+      "epoch": 98.68,
+      "learning_rate": 8.89908256880734e-05,
+      "loss": 0.1243,
+      "step": 10756
+    },
+    {
+      "epoch": 98.69,
+      "learning_rate": 8.895259938837922e-05,
+      "loss": 0.0912,
+      "step": 10757
+    },
+    {
+      "epoch": 98.7,
+      "learning_rate": 8.891437308868501e-05,
+      "loss": 0.1492,
+      "step": 10758
+    },
+    {
+      "epoch": 98.71,
+      "learning_rate": 8.887614678899083e-05,
+      "loss": 0.1033,
+      "step": 10759
+    },
+    {
+      "epoch": 98.72,
+      "learning_rate": 8.883792048929663e-05,
+      "loss": 0.1331,
+      "step": 10760
+    },
+    {
+      "epoch": 98.72,
+      "learning_rate": 8.879969418960245e-05,
+      "loss": 0.094,
+      "step": 10761
+    },
+    {
+      "epoch": 98.73,
+      "learning_rate": 8.876146788990825e-05,
+      "loss": 0.0711,
+      "step": 10762
+    },
+    {
+      "epoch": 98.74,
+      "learning_rate": 8.872324159021407e-05,
+      "loss": 0.0863,
+      "step": 10763
+    },
+    {
+      "epoch": 98.75,
+      "learning_rate": 8.868501529051988e-05,
+      "loss": 0.187,
+      "step": 10764
+    },
+    {
+      "epoch": 98.76,
+      "learning_rate": 8.86467889908257e-05,
+      "loss": 0.1062,
+      "step": 10765
+    },
+    {
+      "epoch": 98.77,
+      "learning_rate": 8.86085626911315e-05,
+      "loss": 0.1589,
+      "step": 10766
+    },
+    {
+      "epoch": 98.78,
+      "learning_rate": 8.857033639143732e-05,
+      "loss": 0.1476,
+      "step": 10767
+    },
+    {
+      "epoch": 98.79,
+      "learning_rate": 8.853211009174312e-05,
+      "loss": 0.1451,
+      "step": 10768
+    },
+    {
+      "epoch": 98.8,
+      "learning_rate": 8.849388379204894e-05,
+      "loss": 0.1703,
+      "step": 10769
+    },
+    {
+      "epoch": 98.81,
+      "learning_rate": 8.845565749235473e-05,
+      "loss": 0.1497,
+      "step": 10770
+    },
+    {
+      "epoch": 98.82,
+      "learning_rate": 8.841743119266055e-05,
+      "loss": 0.0998,
+      "step": 10771
+    },
+    {
+      "epoch": 98.83,
+      "learning_rate": 8.837920489296636e-05,
+      "loss": 0.1499,
+      "step": 10772
+    },
+    {
+      "epoch": 98.83,
+      "learning_rate": 8.834097859327217e-05,
+      "loss": 0.1156,
+      "step": 10773
+    },
+    {
+      "epoch": 98.84,
+      "learning_rate": 8.830275229357799e-05,
+      "loss": 0.1122,
+      "step": 10774
+    },
+    {
+      "epoch": 98.85,
+      "learning_rate": 8.826452599388379e-05,
+      "loss": 0.0793,
+      "step": 10775
+    },
+    {
+      "epoch": 98.86,
+      "learning_rate": 8.822629969418961e-05,
+      "loss": 0.122,
+      "step": 10776
+    },
+    {
+      "epoch": 98.87,
+      "learning_rate": 8.818807339449541e-05,
+      "loss": 0.1644,
+      "step": 10777
+    },
+    {
+      "epoch": 98.88,
+      "learning_rate": 8.814984709480123e-05,
+      "loss": 0.1835,
+      "step": 10778
+    },
+    {
+      "epoch": 98.89,
+      "learning_rate": 8.811162079510704e-05,
+      "loss": 0.0807,
+      "step": 10779
+    },
+    {
+      "epoch": 98.9,
+      "learning_rate": 8.807339449541285e-05,
+      "loss": 0.1641,
+      "step": 10780
+    },
+    {
+      "epoch": 98.91,
+      "learning_rate": 8.803516819571866e-05,
+      "loss": 0.0923,
+      "step": 10781
+    },
+    {
+      "epoch": 98.92,
+      "learning_rate": 8.799694189602446e-05,
+      "loss": 0.1026,
+      "step": 10782
+    },
+    {
+      "epoch": 98.93,
+      "learning_rate": 8.795871559633027e-05,
+      "loss": 0.1223,
+      "step": 10783
+    },
+    {
+      "epoch": 98.94,
+      "learning_rate": 8.792048929663608e-05,
+      "loss": 0.1044,
+      "step": 10784
+    },
+    {
+      "epoch": 98.94,
+      "learning_rate": 8.788226299694189e-05,
+      "loss": 0.2322,
+      "step": 10785
+    },
+    {
+      "epoch": 98.95,
+      "learning_rate": 8.78440366972477e-05,
+      "loss": 0.06,
+      "step": 10786
+    },
+    {
+      "epoch": 98.96,
+      "learning_rate": 8.780581039755352e-05,
+      "loss": 0.102,
+      "step": 10787
+    },
+    {
+      "epoch": 98.97,
+      "learning_rate": 8.776758409785933e-05,
+      "loss": 0.1075,
+      "step": 10788
+    },
+    {
+      "epoch": 98.98,
+      "learning_rate": 8.772935779816515e-05,
+      "loss": 0.1429,
+      "step": 10789
+    },
+    {
+      "epoch": 98.99,
+      "learning_rate": 8.769113149847095e-05,
+      "loss": 0.0324,
+      "step": 10790
+    },
+    {
+      "epoch": 99.0,
+      "learning_rate": 8.765290519877677e-05,
+      "loss": 0.1515,
+      "step": 10791
+    },
+    {
+      "epoch": 99.01,
+      "learning_rate": 8.761467889908257e-05,
+      "loss": 0.2012,
+      "step": 10792
+    },
+    {
+      "epoch": 99.02,
+      "learning_rate": 8.757645259938839e-05,
+      "loss": 0.1535,
+      "step": 10793
+    },
+    {
+      "epoch": 99.03,
+      "learning_rate": 8.753822629969418e-05,
+      "loss": 0.137,
+      "step": 10794
+    },
+    {
+      "epoch": 99.04,
+      "learning_rate": 8.75e-05,
+      "loss": 0.143,
+      "step": 10795
+    },
+    {
+      "epoch": 99.05,
+      "learning_rate": 8.74617737003058e-05,
+      "loss": 0.1264,
+      "step": 10796
+    },
+    {
+      "epoch": 99.06,
+      "learning_rate": 8.742354740061162e-05,
+      "loss": 0.1716,
+      "step": 10797
+    },
+    {
+      "epoch": 99.06,
+      "learning_rate": 8.738532110091743e-05,
+      "loss": 0.1515,
+      "step": 10798
+    },
+    {
+      "epoch": 99.07,
+      "learning_rate": 8.734709480122324e-05,
+      "loss": 0.1232,
+      "step": 10799
+    },
+    {
+      "epoch": 99.08,
+      "learning_rate": 8.730886850152905e-05,
+      "loss": 0.1492,
+      "step": 10800
+    },
+    {
+      "epoch": 99.09,
+      "learning_rate": 8.727064220183487e-05,
+      "loss": 0.0875,
+      "step": 10801
+    },
+    {
+      "epoch": 99.1,
+      "learning_rate": 8.723241590214068e-05,
+      "loss": 0.16,
+      "step": 10802
+    },
+    {
+      "epoch": 99.11,
+      "learning_rate": 8.719418960244649e-05,
+      "loss": 0.1293,
+      "step": 10803
+    },
+    {
+      "epoch": 99.12,
+      "learning_rate": 8.71559633027523e-05,
+      "loss": 0.0972,
+      "step": 10804
+    },
+    {
+      "epoch": 99.13,
+      "learning_rate": 8.711773700305811e-05,
+      "loss": 0.1757,
+      "step": 10805
+    },
+    {
+      "epoch": 99.14,
+      "learning_rate": 8.707951070336391e-05,
+      "loss": 0.1331,
+      "step": 10806
+    },
+    {
+      "epoch": 99.15,
+      "learning_rate": 8.704128440366972e-05,
+      "loss": 0.1428,
+      "step": 10807
+    },
+    {
+      "epoch": 99.16,
+      "learning_rate": 8.700305810397554e-05,
+      "loss": 0.0453,
+      "step": 10808
+    },
+    {
+      "epoch": 99.17,
+      "learning_rate": 8.696483180428134e-05,
+      "loss": 0.147,
+      "step": 10809
+    },
+    {
+      "epoch": 99.17,
+      "learning_rate": 8.692660550458716e-05,
+      "loss": 0.1011,
+      "step": 10810
+    },
+    {
+      "epoch": 99.18,
+      "learning_rate": 8.688837920489296e-05,
+      "loss": 0.138,
+      "step": 10811
+    },
+    {
+      "epoch": 99.19,
+      "learning_rate": 8.685015290519878e-05,
+      "loss": 0.1202,
+      "step": 10812
+    },
+    {
+      "epoch": 99.2,
+      "learning_rate": 8.681192660550458e-05,
+      "loss": 0.156,
+      "step": 10813
+    },
+    {
+      "epoch": 99.21,
+      "learning_rate": 8.67737003058104e-05,
+      "loss": 0.0772,
+      "step": 10814
+    },
+    {
+      "epoch": 99.22,
+      "learning_rate": 8.673547400611621e-05,
+      "loss": 0.143,
+      "step": 10815
+    },
+    {
+      "epoch": 99.23,
+      "learning_rate": 8.669724770642202e-05,
+      "loss": 0.0994,
+      "step": 10816
+    },
+    {
+      "epoch": 99.24,
+      "learning_rate": 8.665902140672784e-05,
+      "loss": 0.129,
+      "step": 10817
+    },
+    {
+      "epoch": 99.25,
+      "learning_rate": 8.662079510703363e-05,
+      "loss": 0.087,
+      "step": 10818
+    },
+    {
+      "epoch": 99.26,
+      "learning_rate": 8.658256880733945e-05,
+      "loss": 0.1696,
+      "step": 10819
+    },
+    {
+      "epoch": 99.27,
+      "learning_rate": 8.654434250764526e-05,
+      "loss": 0.1857,
+      "step": 10820
+    },
+    {
+      "epoch": 99.28,
+      "learning_rate": 8.650611620795107e-05,
+      "loss": 0.1341,
+      "step": 10821
+    },
+    {
+      "epoch": 99.28,
+      "learning_rate": 8.646788990825688e-05,
+      "loss": 0.139,
+      "step": 10822
+    },
+    {
+      "epoch": 99.29,
+      "learning_rate": 8.64296636085627e-05,
+      "loss": 0.1224,
+      "step": 10823
+    },
+    {
+      "epoch": 99.3,
+      "learning_rate": 8.63914373088685e-05,
+      "loss": 0.135,
+      "step": 10824
+    },
+    {
+      "epoch": 99.31,
+      "learning_rate": 8.635321100917432e-05,
+      "loss": 0.1548,
+      "step": 10825
+    },
+    {
+      "epoch": 99.32,
+      "learning_rate": 8.631498470948012e-05,
+      "loss": 0.1271,
+      "step": 10826
+    },
+    {
+      "epoch": 99.33,
+      "learning_rate": 8.627675840978594e-05,
+      "loss": 0.1563,
+      "step": 10827
+    },
+    {
+      "epoch": 99.34,
+      "learning_rate": 8.623853211009174e-05,
+      "loss": 0.0791,
+      "step": 10828
+    },
+    {
+      "epoch": 99.35,
+      "learning_rate": 8.620030581039756e-05,
+      "loss": 0.1151,
+      "step": 10829
+    },
+    {
+      "epoch": 99.36,
+      "learning_rate": 8.616207951070335e-05,
+      "loss": 0.0829,
+      "step": 10830
+    },
+    {
+      "epoch": 99.37,
+      "learning_rate": 8.612385321100917e-05,
+      "loss": 0.0801,
+      "step": 10831
+    },
+    {
+      "epoch": 99.38,
+      "learning_rate": 8.608562691131499e-05,
+      "loss": 0.1483,
+      "step": 10832
+    },
+    {
+      "epoch": 99.39,
+      "learning_rate": 8.604740061162079e-05,
+      "loss": 0.17,
+      "step": 10833
+    },
+    {
+      "epoch": 99.39,
+      "learning_rate": 8.600917431192661e-05,
+      "loss": 0.1207,
+      "step": 10834
+    },
+    {
+      "epoch": 99.4,
+      "learning_rate": 8.597094801223241e-05,
+      "loss": 0.0917,
+      "step": 10835
+    },
+    {
+      "epoch": 99.41,
+      "learning_rate": 8.593272171253823e-05,
+      "loss": 0.1287,
+      "step": 10836
+    },
+    {
+      "epoch": 99.42,
+      "learning_rate": 8.589449541284404e-05,
+      "loss": 0.107,
+      "step": 10837
+    },
+    {
+      "epoch": 99.43,
+      "learning_rate": 8.585626911314985e-05,
+      "loss": 0.0868,
+      "step": 10838
+    },
+    {
+      "epoch": 99.44,
+      "learning_rate": 8.581804281345566e-05,
+      "loss": 0.094,
+      "step": 10839
+    },
+    {
+      "epoch": 99.45,
+      "learning_rate": 8.577981651376148e-05,
+      "loss": 0.0835,
+      "step": 10840
+    },
+    {
+      "epoch": 99.46,
+      "learning_rate": 8.574159021406728e-05,
+      "loss": 0.0495,
+      "step": 10841
+    },
+    {
+      "epoch": 99.47,
+      "learning_rate": 8.570336391437309e-05,
+      "loss": 0.104,
+      "step": 10842
+    },
+    {
+      "epoch": 99.48,
+      "learning_rate": 8.566513761467889e-05,
+      "loss": 0.1175,
+      "step": 10843
+    },
+    {
+      "epoch": 99.49,
+      "learning_rate": 8.562691131498471e-05,
+      "loss": 0.1384,
+      "step": 10844
+    },
+    {
+      "epoch": 99.5,
+      "learning_rate": 8.558868501529051e-05,
+      "loss": 0.0741,
+      "step": 10845
+    },
+    {
+      "epoch": 99.5,
+      "learning_rate": 8.555045871559633e-05,
+      "loss": 0.2468,
+      "step": 10846
+    },
+    {
+      "epoch": 99.51,
+      "learning_rate": 8.551223241590215e-05,
+      "loss": 0.1536,
+      "step": 10847
+    },
+    {
+      "epoch": 99.52,
+      "learning_rate": 8.547400611620795e-05,
+      "loss": 0.1348,
+      "step": 10848
+    },
+    {
+      "epoch": 99.53,
+      "learning_rate": 8.543577981651377e-05,
+      "loss": 0.1301,
+      "step": 10849
+    },
+    {
+      "epoch": 99.54,
+      "learning_rate": 8.539755351681957e-05,
+      "loss": 0.1523,
+      "step": 10850
+    },
+    {
+      "epoch": 99.55,
+      "learning_rate": 8.535932721712539e-05,
+      "loss": 0.0997,
+      "step": 10851
+    },
+    {
+      "epoch": 99.56,
+      "learning_rate": 8.53211009174312e-05,
+      "loss": 0.1762,
+      "step": 10852
+    },
+    {
+      "epoch": 99.57,
+      "learning_rate": 8.528287461773701e-05,
+      "loss": 0.113,
+      "step": 10853
+    },
+    {
+      "epoch": 99.58,
+      "learning_rate": 8.52446483180428e-05,
+      "loss": 0.141,
+      "step": 10854
+    },
+    {
+      "epoch": 99.59,
+      "learning_rate": 8.520642201834862e-05,
+      "loss": 0.1972,
+      "step": 10855
+    },
+    {
+      "epoch": 99.6,
+      "learning_rate": 8.516819571865443e-05,
+      "loss": 0.0946,
+      "step": 10856
+    },
+    {
+      "epoch": 99.61,
+      "learning_rate": 8.512996941896024e-05,
+      "loss": 0.1621,
+      "step": 10857
+    },
+    {
+      "epoch": 99.61,
+      "learning_rate": 8.509174311926605e-05,
+      "loss": 0.086,
+      "step": 10858
+    },
+    {
+      "epoch": 99.62,
+      "learning_rate": 8.505351681957187e-05,
+      "loss": 0.1551,
+      "step": 10859
+    },
+    {
+      "epoch": 99.63,
+      "learning_rate": 8.501529051987768e-05,
+      "loss": 0.1505,
+      "step": 10860
+    },
+    {
+      "epoch": 99.64,
+      "learning_rate": 8.497706422018349e-05,
+      "loss": 0.1204,
+      "step": 10861
+    },
+    {
+      "epoch": 99.65,
+      "learning_rate": 8.49388379204893e-05,
+      "loss": 0.0843,
+      "step": 10862
+    },
+    {
+      "epoch": 99.66,
+      "learning_rate": 8.490061162079511e-05,
+      "loss": 0.0816,
+      "step": 10863
+    },
+    {
+      "epoch": 99.67,
+      "learning_rate": 8.486238532110093e-05,
+      "loss": 0.1079,
+      "step": 10864
+    },
+    {
+      "epoch": 99.68,
+      "learning_rate": 8.482415902140673e-05,
+      "loss": 0.1357,
+      "step": 10865
+    },
+    {
+      "epoch": 99.69,
+      "learning_rate": 8.478593272171254e-05,
+      "loss": 0.0843,
+      "step": 10866
+    },
+    {
+      "epoch": 99.7,
+      "learning_rate": 8.474770642201834e-05,
+      "loss": 0.1788,
+      "step": 10867
+    },
+    {
+      "epoch": 99.71,
+      "learning_rate": 8.470948012232416e-05,
+      "loss": 0.084,
+      "step": 10868
+    },
+    {
+      "epoch": 99.72,
+      "learning_rate": 8.467125382262996e-05,
+      "loss": 0.1195,
+      "step": 10869
+    },
+    {
+      "epoch": 99.72,
+      "learning_rate": 8.463302752293578e-05,
+      "loss": 0.0697,
+      "step": 10870
+    },
+    {
+      "epoch": 99.73,
+      "learning_rate": 8.459480122324159e-05,
+      "loss": 0.0917,
+      "step": 10871
+    },
+    {
+      "epoch": 99.74,
+      "learning_rate": 8.45565749235474e-05,
+      "loss": 0.2404,
+      "step": 10872
+    },
+    {
+      "epoch": 99.75,
+      "learning_rate": 8.451834862385321e-05,
+      "loss": 0.1657,
+      "step": 10873
+    },
+    {
+      "epoch": 99.76,
+      "learning_rate": 8.448012232415903e-05,
+      "loss": 0.1577,
+      "step": 10874
+    },
+    {
+      "epoch": 99.77,
+      "learning_rate": 8.444189602446484e-05,
+      "loss": 0.1542,
+      "step": 10875
+    },
+    {
+      "epoch": 99.78,
+      "learning_rate": 8.440366972477065e-05,
+      "loss": 0.1136,
+      "step": 10876
+    },
+    {
+      "epoch": 99.79,
+      "learning_rate": 8.436544342507647e-05,
+      "loss": 0.2198,
+      "step": 10877
+    },
+    {
+      "epoch": 99.8,
+      "learning_rate": 8.432721712538226e-05,
+      "loss": 0.1766,
+      "step": 10878
+    },
+    {
+      "epoch": 99.81,
+      "learning_rate": 8.428899082568807e-05,
+      "loss": 0.092,
+      "step": 10879
+    },
+    {
+      "epoch": 99.82,
+      "learning_rate": 8.425076452599388e-05,
+      "loss": 0.1227,
+      "step": 10880
+    },
+    {
+      "epoch": 99.83,
+      "learning_rate": 8.42125382262997e-05,
+      "loss": 0.1639,
+      "step": 10881
+    },
+    {
+      "epoch": 99.83,
+      "learning_rate": 8.41743119266055e-05,
+      "loss": 0.1598,
+      "step": 10882
+    },
+    {
+      "epoch": 99.84,
+      "learning_rate": 8.413608562691132e-05,
+      "loss": 0.1785,
+      "step": 10883
+    },
+    {
+      "epoch": 99.85,
+      "learning_rate": 8.409785932721712e-05,
+      "loss": 0.1099,
+      "step": 10884
+    },
+    {
+      "epoch": 99.86,
+      "learning_rate": 8.405963302752294e-05,
+      "loss": 0.1522,
+      "step": 10885
+    },
+    {
+      "epoch": 99.87,
+      "learning_rate": 8.402140672782874e-05,
+      "loss": 0.1384,
+      "step": 10886
+    },
+    {
+      "epoch": 99.88,
+      "learning_rate": 8.398318042813456e-05,
+      "loss": 0.1241,
+      "step": 10887
+    },
+    {
+      "epoch": 99.89,
+      "learning_rate": 8.394495412844037e-05,
+      "loss": 0.1853,
+      "step": 10888
+    },
+    {
+      "epoch": 99.9,
+      "learning_rate": 8.390672782874618e-05,
+      "loss": 0.1023,
+      "step": 10889
+    },
+    {
+      "epoch": 99.91,
+      "learning_rate": 8.386850152905199e-05,
+      "loss": 0.1009,
+      "step": 10890
+    },
+    {
+      "epoch": 99.92,
+      "learning_rate": 8.38302752293578e-05,
+      "loss": 0.1259,
+      "step": 10891
+    },
+    {
+      "epoch": 99.93,
+      "learning_rate": 8.379204892966361e-05,
+      "loss": 0.1039,
+      "step": 10892
+    },
+    {
+      "epoch": 99.94,
+      "learning_rate": 8.375382262996942e-05,
+      "loss": 0.0836,
+      "step": 10893
+    },
+    {
+      "epoch": 99.94,
+      "learning_rate": 8.371559633027523e-05,
+      "loss": 0.1303,
+      "step": 10894
+    },
+    {
+      "epoch": 99.95,
+      "learning_rate": 8.367737003058104e-05,
+      "loss": 0.0629,
+      "step": 10895
+    },
+    {
+      "epoch": 99.96,
+      "learning_rate": 8.363914373088686e-05,
+      "loss": 0.1409,
+      "step": 10896
+    },
+    {
+      "epoch": 99.97,
+      "learning_rate": 8.360091743119266e-05,
+      "loss": 0.1058,
+      "step": 10897
+    },
+    {
+      "epoch": 99.98,
+      "learning_rate": 8.356269113149848e-05,
+      "loss": 0.1542,
+      "step": 10898
+    },
+    {
+      "epoch": 99.99,
+      "learning_rate": 8.352446483180428e-05,
+      "loss": 0.1056,
+      "step": 10899
+    },
+    {
+      "epoch": 100.0,
+      "learning_rate": 8.34862385321101e-05,
+      "loss": 0.1382,
+      "step": 10900
+    },
+    {
+      "epoch": 100.01,
+      "learning_rate": 8.34480122324159e-05,
+      "loss": 0.2297,
+      "step": 10901
+    },
+    {
+      "epoch": 100.02,
+      "learning_rate": 8.340978593272171e-05,
+      "loss": 0.176,
+      "step": 10902
+    },
+    {
+      "epoch": 100.03,
+      "learning_rate": 8.337155963302751e-05,
+      "loss": 0.1407,
+      "step": 10903
+    },
+    {
+      "epoch": 100.04,
+      "learning_rate": 8.333333333333333e-05,
+      "loss": 0.0956,
+      "step": 10904
+    },
+    {
+      "epoch": 100.05,
+      "learning_rate": 8.329510703363915e-05,
+      "loss": 0.1639,
+      "step": 10905
+    },
+    {
+      "epoch": 100.06,
+      "learning_rate": 8.325688073394495e-05,
+      "loss": 0.1244,
+      "step": 10906
+    },
+    {
+      "epoch": 100.06,
+      "learning_rate": 8.321865443425077e-05,
+      "loss": 0.1043,
+      "step": 10907
+    },
+    {
+      "epoch": 100.07,
+      "learning_rate": 8.318042813455657e-05,
+      "loss": 0.1223,
+      "step": 10908
+    },
+    {
+      "epoch": 100.08,
+      "learning_rate": 8.314220183486239e-05,
+      "loss": 0.1644,
+      "step": 10909
+    },
+    {
+      "epoch": 100.09,
+      "learning_rate": 8.31039755351682e-05,
+      "loss": 0.0975,
+      "step": 10910
+    },
+    {
+      "epoch": 100.1,
+      "learning_rate": 8.306574923547401e-05,
+      "loss": 0.092,
+      "step": 10911
+    },
+    {
+      "epoch": 100.11,
+      "learning_rate": 8.302752293577982e-05,
+      "loss": 0.1564,
+      "step": 10912
+    },
+    {
+      "epoch": 100.12,
+      "learning_rate": 8.298929663608564e-05,
+      "loss": 0.1079,
+      "step": 10913
+    },
+    {
+      "epoch": 100.13,
+      "learning_rate": 8.295107033639144e-05,
+      "loss": 0.0921,
+      "step": 10914
+    },
+    {
+      "epoch": 100.14,
+      "learning_rate": 8.291284403669725e-05,
+      "loss": 0.0912,
+      "step": 10915
+    },
+    {
+      "epoch": 100.15,
+      "learning_rate": 8.287461773700305e-05,
+      "loss": 0.1305,
+      "step": 10916
+    },
+    {
+      "epoch": 100.16,
+      "learning_rate": 8.283639143730887e-05,
+      "loss": 0.0927,
+      "step": 10917
+    },
+    {
+      "epoch": 100.17,
+      "learning_rate": 8.279816513761467e-05,
+      "loss": 0.139,
+      "step": 10918
+    },
+    {
+      "epoch": 100.17,
+      "learning_rate": 8.275993883792049e-05,
+      "loss": 0.1169,
+      "step": 10919
+    },
+    {
+      "epoch": 100.18,
+      "learning_rate": 8.272171253822631e-05,
+      "loss": 0.1092,
+      "step": 10920
+    },
+    {
+      "epoch": 100.19,
+      "learning_rate": 8.268348623853211e-05,
+      "loss": 0.0988,
+      "step": 10921
+    },
+    {
+      "epoch": 100.2,
+      "learning_rate": 8.264525993883793e-05,
+      "loss": 0.1263,
+      "step": 10922
+    },
+    {
+      "epoch": 100.21,
+      "learning_rate": 8.260703363914373e-05,
+      "loss": 0.1033,
+      "step": 10923
+    },
+    {
+      "epoch": 100.22,
+      "learning_rate": 8.256880733944955e-05,
+      "loss": 0.1157,
+      "step": 10924
+    },
+    {
+      "epoch": 100.23,
+      "learning_rate": 8.253058103975536e-05,
+      "loss": 0.086,
+      "step": 10925
+    },
+    {
+      "epoch": 100.24,
+      "learning_rate": 8.249235474006117e-05,
+      "loss": 0.0727,
+      "step": 10926
+    },
+    {
+      "epoch": 100.25,
+      "learning_rate": 8.245412844036696e-05,
+      "loss": 0.0329,
+      "step": 10927
+    },
+    {
+      "epoch": 100.26,
+      "learning_rate": 8.241590214067278e-05,
+      "loss": 0.1893,
+      "step": 10928
+    },
+    {
+      "epoch": 100.27,
+      "learning_rate": 8.237767584097859e-05,
+      "loss": 0.1486,
+      "step": 10929
+    },
+    {
+      "epoch": 100.28,
+      "learning_rate": 8.23394495412844e-05,
+      "loss": 0.1911,
+      "step": 10930
+    },
+    {
+      "epoch": 100.28,
+      "learning_rate": 8.230122324159021e-05,
+      "loss": 0.112,
+      "step": 10931
+    },
+    {
+      "epoch": 100.29,
+      "learning_rate": 8.226299694189603e-05,
+      "loss": 0.1417,
+      "step": 10932
+    },
+    {
+      "epoch": 100.3,
+      "learning_rate": 8.222477064220183e-05,
+      "loss": 0.1225,
+      "step": 10933
+    },
+    {
+      "epoch": 100.31,
+      "learning_rate": 8.218654434250765e-05,
+      "loss": 0.1323,
+      "step": 10934
+    },
+    {
+      "epoch": 100.32,
+      "learning_rate": 8.214831804281347e-05,
+      "loss": 0.1809,
+      "step": 10935
+    },
+    {
+      "epoch": 100.33,
+      "learning_rate": 8.211009174311927e-05,
+      "loss": 0.1204,
+      "step": 10936
+    },
+    {
+      "epoch": 100.34,
+      "learning_rate": 8.207186544342509e-05,
+      "loss": 0.1507,
+      "step": 10937
+    },
+    {
+      "epoch": 100.35,
+      "learning_rate": 8.203363914373089e-05,
+      "loss": 0.0865,
+      "step": 10938
+    },
+    {
+      "epoch": 100.36,
+      "learning_rate": 8.19954128440367e-05,
+      "loss": 0.1586,
+      "step": 10939
+    },
+    {
+      "epoch": 100.37,
+      "learning_rate": 8.19571865443425e-05,
+      "loss": 0.1893,
+      "step": 10940
+    },
+    {
+      "epoch": 100.38,
+      "learning_rate": 8.191896024464832e-05,
+      "loss": 0.1218,
+      "step": 10941
+    },
+    {
+      "epoch": 100.39,
+      "learning_rate": 8.188073394495412e-05,
+      "loss": 0.1345,
+      "step": 10942
+    },
+    {
+      "epoch": 100.39,
+      "learning_rate": 8.184250764525994e-05,
+      "loss": 0.1171,
+      "step": 10943
+    },
+    {
+      "epoch": 100.4,
+      "learning_rate": 8.180428134556575e-05,
+      "loss": 0.1258,
+      "step": 10944
+    },
+    {
+      "epoch": 100.41,
+      "learning_rate": 8.176605504587156e-05,
+      "loss": 0.0827,
+      "step": 10945
+    },
+    {
+      "epoch": 100.42,
+      "learning_rate": 8.172782874617737e-05,
+      "loss": 0.1376,
+      "step": 10946
+    },
+    {
+      "epoch": 100.43,
+      "learning_rate": 8.168960244648319e-05,
+      "loss": 0.1663,
+      "step": 10947
+    },
+    {
+      "epoch": 100.44,
+      "learning_rate": 8.1651376146789e-05,
+      "loss": 0.119,
+      "step": 10948
+    },
+    {
+      "epoch": 100.45,
+      "learning_rate": 8.161314984709481e-05,
+      "loss": 0.0682,
+      "step": 10949
+    },
+    {
+      "epoch": 100.46,
+      "learning_rate": 8.157492354740063e-05,
+      "loss": 0.0879,
+      "step": 10950
+    },
+    {
+      "epoch": 100.47,
+      "learning_rate": 8.153669724770642e-05,
+      "loss": 0.1122,
+      "step": 10951
+    },
+    {
+      "epoch": 100.48,
+      "learning_rate": 8.149847094801223e-05,
+      "loss": 0.1447,
+      "step": 10952
+    },
+    {
+      "epoch": 100.49,
+      "learning_rate": 8.146024464831804e-05,
+      "loss": 0.1115,
+      "step": 10953
+    },
+    {
+      "epoch": 100.5,
+      "learning_rate": 8.142201834862386e-05,
+      "loss": 0.0876,
+      "step": 10954
+    },
+    {
+      "epoch": 100.5,
+      "learning_rate": 8.138379204892966e-05,
+      "loss": 0.2055,
+      "step": 10955
+    },
+    {
+      "epoch": 100.51,
+      "learning_rate": 8.134556574923548e-05,
+      "loss": 0.1742,
+      "step": 10956
+    },
+    {
+      "epoch": 100.52,
+      "learning_rate": 8.130733944954128e-05,
+      "loss": 0.1612,
+      "step": 10957
+    },
+    {
+      "epoch": 100.53,
+      "learning_rate": 8.12691131498471e-05,
+      "loss": 0.1215,
+      "step": 10958
+    },
+    {
+      "epoch": 100.54,
+      "learning_rate": 8.12308868501529e-05,
+      "loss": 0.1598,
+      "step": 10959
+    },
+    {
+      "epoch": 100.55,
+      "learning_rate": 8.119266055045872e-05,
+      "loss": 0.0829,
+      "step": 10960
+    },
+    {
+      "epoch": 100.56,
+      "learning_rate": 8.115443425076453e-05,
+      "loss": 0.1668,
+      "step": 10961
+    },
+    {
+      "epoch": 100.57,
+      "learning_rate": 8.111620795107035e-05,
+      "loss": 0.0964,
+      "step": 10962
+    },
+    {
+      "epoch": 100.58,
+      "learning_rate": 8.107798165137614e-05,
+      "loss": 0.1044,
+      "step": 10963
+    },
+    {
+      "epoch": 100.59,
+      "learning_rate": 8.103975535168195e-05,
+      "loss": 0.0917,
+      "step": 10964
+    },
+    {
+      "epoch": 100.6,
+      "learning_rate": 8.100152905198777e-05,
+      "loss": 0.0911,
+      "step": 10965
+    },
+    {
+      "epoch": 100.61,
+      "learning_rate": 8.096330275229358e-05,
+      "loss": 0.0838,
+      "step": 10966
+    },
+    {
+      "epoch": 100.61,
+      "learning_rate": 8.09250764525994e-05,
+      "loss": 0.1171,
+      "step": 10967
+    },
+    {
+      "epoch": 100.62,
+      "learning_rate": 8.08868501529052e-05,
+      "loss": 0.1525,
+      "step": 10968
+    },
+    {
+      "epoch": 100.63,
+      "learning_rate": 8.084862385321102e-05,
+      "loss": 0.0612,
+      "step": 10969
+    },
+    {
+      "epoch": 100.64,
+      "learning_rate": 8.081039755351682e-05,
+      "loss": 0.0533,
+      "step": 10970
+    },
+    {
+      "epoch": 100.65,
+      "learning_rate": 8.077217125382264e-05,
+      "loss": 0.1138,
+      "step": 10971
+    },
+    {
+      "epoch": 100.66,
+      "learning_rate": 8.073394495412844e-05,
+      "loss": 0.111,
+      "step": 10972
+    },
+    {
+      "epoch": 100.67,
+      "learning_rate": 8.069571865443426e-05,
+      "loss": 0.0599,
+      "step": 10973
+    },
+    {
+      "epoch": 100.68,
+      "learning_rate": 8.065749235474006e-05,
+      "loss": 0.1704,
+      "step": 10974
+    },
+    {
+      "epoch": 100.69,
+      "learning_rate": 8.061926605504587e-05,
+      "loss": 0.1029,
+      "step": 10975
+    },
+    {
+      "epoch": 100.7,
+      "learning_rate": 8.058103975535167e-05,
+      "loss": 0.1379,
+      "step": 10976
+    },
+    {
+      "epoch": 100.71,
+      "learning_rate": 8.054281345565749e-05,
+      "loss": 0.0974,
+      "step": 10977
+    },
+    {
+      "epoch": 100.72,
+      "learning_rate": 8.050458715596331e-05,
+      "loss": 0.1202,
+      "step": 10978
+    },
+    {
+      "epoch": 100.72,
+      "learning_rate": 8.046636085626911e-05,
+      "loss": 0.1289,
+      "step": 10979
+    },
+    {
+      "epoch": 100.73,
+      "learning_rate": 8.042813455657493e-05,
+      "loss": 0.0734,
+      "step": 10980
+    },
+    {
+      "epoch": 100.74,
+      "learning_rate": 8.038990825688074e-05,
+      "loss": 0.1213,
+      "step": 10981
+    },
+    {
+      "epoch": 100.75,
+      "learning_rate": 8.035168195718655e-05,
+      "loss": 0.176,
+      "step": 10982
+    },
+    {
+      "epoch": 100.76,
+      "learning_rate": 8.031345565749236e-05,
+      "loss": 0.1457,
+      "step": 10983
+    },
+    {
+      "epoch": 100.77,
+      "learning_rate": 8.027522935779818e-05,
+      "loss": 0.1069,
+      "step": 10984
+    },
+    {
+      "epoch": 100.78,
+      "learning_rate": 8.023700305810398e-05,
+      "loss": 0.1208,
+      "step": 10985
+    },
+    {
+      "epoch": 100.79,
+      "learning_rate": 8.01987767584098e-05,
+      "loss": 0.2068,
+      "step": 10986
+    },
+    {
+      "epoch": 100.8,
+      "learning_rate": 8.016055045871559e-05,
+      "loss": 0.0944,
+      "step": 10987
+    },
+    {
+      "epoch": 100.81,
+      "learning_rate": 8.01223241590214e-05,
+      "loss": 0.0853,
+      "step": 10988
+    },
+    {
+      "epoch": 100.82,
+      "learning_rate": 8.008409785932721e-05,
+      "loss": 0.1,
+      "step": 10989
+    },
+    {
+      "epoch": 100.83,
+      "learning_rate": 8.004587155963303e-05,
+      "loss": 0.0953,
+      "step": 10990
+    },
+    {
+      "epoch": 100.83,
+      "learning_rate": 8.000764525993883e-05,
+      "loss": 0.1048,
+      "step": 10991
+    },
+    {
+      "epoch": 100.84,
+      "learning_rate": 7.996941896024465e-05,
+      "loss": 0.1213,
+      "step": 10992
+    },
+    {
+      "epoch": 100.85,
+      "learning_rate": 7.993119266055047e-05,
+      "loss": 0.1279,
+      "step": 10993
+    },
+    {
+      "epoch": 100.86,
+      "learning_rate": 7.989296636085627e-05,
+      "loss": 0.1786,
+      "step": 10994
+    },
+    {
+      "epoch": 100.87,
+      "learning_rate": 7.985474006116209e-05,
+      "loss": 0.1107,
+      "step": 10995
+    },
+    {
+      "epoch": 100.88,
+      "learning_rate": 7.98165137614679e-05,
+      "loss": 0.1014,
+      "step": 10996
+    },
+    {
+      "epoch": 100.89,
+      "learning_rate": 7.977828746177371e-05,
+      "loss": 0.1331,
+      "step": 10997
+    },
+    {
+      "epoch": 100.9,
+      "learning_rate": 7.974006116207952e-05,
+      "loss": 0.1205,
+      "step": 10998
+    },
+    {
+      "epoch": 100.91,
+      "learning_rate": 7.970183486238532e-05,
+      "loss": 0.1238,
+      "step": 10999
+    },
+    {
+      "epoch": 100.92,
+      "learning_rate": 7.966360856269112e-05,
+      "loss": 0.0896,
+      "step": 11000
+    },
+    {
+      "epoch": 100.92,
+      "eval_cer": 0.1398071935111471,
+      "eval_loss": 0.7186796069145203,
+      "eval_runtime": 87.2793,
+      "eval_samples_per_second": 18.87,
+      "eval_steps_per_second": 2.36,
+      "eval_wer": 0.4885098559901951,
+      "step": 11000
+    },
+    {
+      "epoch": 100.93,
+      "learning_rate": 7.962538226299694e-05,
+      "loss": 0.0813,
+      "step": 11001
+    },
+    {
+      "epoch": 100.94,
+      "learning_rate": 7.958715596330275e-05,
+      "loss": 0.0822,
+      "step": 11002
+    },
+    {
+      "epoch": 100.94,
+      "learning_rate": 7.954892966360856e-05,
+      "loss": 0.1016,
+      "step": 11003
+    },
+    {
+      "epoch": 100.95,
+      "learning_rate": 7.951070336391437e-05,
+      "loss": 0.1111,
+      "step": 11004
+    },
+    {
+      "epoch": 100.96,
+      "learning_rate": 7.947247706422019e-05,
+      "loss": 0.0882,
+      "step": 11005
+    },
+    {
+      "epoch": 100.97,
+      "learning_rate": 7.943425076452599e-05,
+      "loss": 0.1406,
+      "step": 11006
+    },
+    {
+      "epoch": 100.98,
+      "learning_rate": 7.939602446483181e-05,
+      "loss": 0.0808,
+      "step": 11007
+    },
+    {
+      "epoch": 100.99,
+      "learning_rate": 7.935779816513763e-05,
+      "loss": 0.083,
+      "step": 11008
+    },
+    {
+      "epoch": 101.0,
+      "learning_rate": 7.931957186544343e-05,
+      "loss": 0.1158,
+      "step": 11009
+    },
+    {
+      "epoch": 101.01,
+      "learning_rate": 7.928134556574925e-05,
+      "loss": 0.1394,
+      "step": 11010
+    },
+    {
+      "epoch": 101.02,
+      "learning_rate": 7.924311926605504e-05,
+      "loss": 0.1347,
+      "step": 11011
+    },
+    {
+      "epoch": 101.03,
+      "learning_rate": 7.920489296636086e-05,
+      "loss": 0.1475,
+      "step": 11012
+    },
+    {
+      "epoch": 101.04,
+      "learning_rate": 7.916666666666666e-05,
+      "loss": 0.165,
+      "step": 11013
+    },
+    {
+      "epoch": 101.05,
+      "learning_rate": 7.912844036697248e-05,
+      "loss": 0.1418,
+      "step": 11014
+    },
+    {
+      "epoch": 101.06,
+      "learning_rate": 7.909021406727828e-05,
+      "loss": 0.0649,
+      "step": 11015
+    },
+    {
+      "epoch": 101.06,
+      "learning_rate": 7.90519877675841e-05,
+      "loss": 0.1036,
+      "step": 11016
+    },
+    {
+      "epoch": 101.07,
+      "learning_rate": 7.90137614678899e-05,
+      "loss": 0.1124,
+      "step": 11017
+    },
+    {
+      "epoch": 101.08,
+      "learning_rate": 7.897553516819572e-05,
+      "loss": 0.1409,
+      "step": 11018
+    },
+    {
+      "epoch": 101.09,
+      "learning_rate": 7.893730886850153e-05,
+      "loss": 0.0802,
+      "step": 11019
+    },
+    {
+      "epoch": 101.1,
+      "learning_rate": 7.889908256880735e-05,
+      "loss": 0.1069,
+      "step": 11020
+    },
+    {
+      "epoch": 101.11,
+      "learning_rate": 7.886085626911315e-05,
+      "loss": 0.1126,
+      "step": 11021
+    },
+    {
+      "epoch": 101.12,
+      "learning_rate": 7.882262996941897e-05,
+      "loss": 0.0992,
+      "step": 11022
+    },
+    {
+      "epoch": 101.13,
+      "learning_rate": 7.878440366972477e-05,
+      "loss": 0.1535,
+      "step": 11023
+    },
+    {
+      "epoch": 101.14,
+      "learning_rate": 7.874617737003058e-05,
+      "loss": 0.1428,
+      "step": 11024
+    },
+    {
+      "epoch": 101.15,
+      "learning_rate": 7.87079510703364e-05,
+      "loss": 0.2055,
+      "step": 11025
+    },
+    {
+      "epoch": 101.16,
+      "learning_rate": 7.86697247706422e-05,
+      "loss": 0.0525,
+      "step": 11026
+    },
+    {
+      "epoch": 101.17,
+      "learning_rate": 7.863149847094802e-05,
+      "loss": 0.1214,
+      "step": 11027
+    },
+    {
+      "epoch": 101.17,
+      "learning_rate": 7.859327217125382e-05,
+      "loss": 0.1075,
+      "step": 11028
+    },
+    {
+      "epoch": 101.18,
+      "learning_rate": 7.855504587155964e-05,
+      "loss": 0.1118,
+      "step": 11029
+    },
+    {
+      "epoch": 101.19,
+      "learning_rate": 7.851681957186544e-05,
+      "loss": 0.1439,
+      "step": 11030
+    },
+    {
+      "epoch": 101.2,
+      "learning_rate": 7.847859327217126e-05,
+      "loss": 0.1298,
+      "step": 11031
+    },
+    {
+      "epoch": 101.21,
+      "learning_rate": 7.844036697247707e-05,
+      "loss": 0.1248,
+      "step": 11032
+    },
+    {
+      "epoch": 101.22,
+      "learning_rate": 7.840214067278288e-05,
+      "loss": 0.078,
+      "step": 11033
+    },
+    {
+      "epoch": 101.23,
+      "learning_rate": 7.836391437308869e-05,
+      "loss": 0.1159,
+      "step": 11034
+    },
+    {
+      "epoch": 101.24,
+      "learning_rate": 7.832568807339449e-05,
+      "loss": 0.1376,
+      "step": 11035
+    },
+    {
+      "epoch": 101.25,
+      "learning_rate": 7.82874617737003e-05,
+      "loss": 0.0324,
+      "step": 11036
+    },
+    {
+      "epoch": 101.26,
+      "learning_rate": 7.824923547400611e-05,
+      "loss": 0.1985,
+      "step": 11037
+    },
+    {
+      "epoch": 101.27,
+      "learning_rate": 7.821100917431193e-05,
+      "loss": 0.1914,
+      "step": 11038
+    },
+    {
+      "epoch": 101.28,
+      "learning_rate": 7.817278287461774e-05,
+      "loss": 0.1351,
+      "step": 11039
+    },
+    {
+      "epoch": 101.28,
+      "learning_rate": 7.813455657492355e-05,
+      "loss": 0.1434,
+      "step": 11040
+    },
+    {
+      "epoch": 101.29,
+      "learning_rate": 7.809633027522936e-05,
+      "loss": 0.1122,
+      "step": 11041
+    },
+    {
+      "epoch": 101.3,
+      "learning_rate": 7.805810397553518e-05,
+      "loss": 0.1489,
+      "step": 11042
+    },
+    {
+      "epoch": 101.31,
+      "learning_rate": 7.801987767584098e-05,
+      "loss": 0.1446,
+      "step": 11043
+    },
+    {
+      "epoch": 101.32,
+      "learning_rate": 7.79816513761468e-05,
+      "loss": 0.144,
+      "step": 11044
+    },
+    {
+      "epoch": 101.33,
+      "learning_rate": 7.79434250764526e-05,
+      "loss": 0.1575,
+      "step": 11045
+    },
+    {
+      "epoch": 101.34,
+      "learning_rate": 7.790519877675842e-05,
+      "loss": 0.1222,
+      "step": 11046
+    },
+    {
+      "epoch": 101.35,
+      "learning_rate": 7.786697247706421e-05,
+      "loss": 0.1185,
+      "step": 11047
+    },
+    {
+      "epoch": 101.36,
+      "learning_rate": 7.782874617737003e-05,
+      "loss": 0.0871,
+      "step": 11048
+    },
+    {
+      "epoch": 101.37,
+      "learning_rate": 7.779051987767583e-05,
+      "loss": 0.1156,
+      "step": 11049
+    },
+    {
+      "epoch": 101.38,
+      "learning_rate": 7.775229357798165e-05,
+      "loss": 0.0691,
+      "step": 11050
+    },
+    {
+      "epoch": 101.39,
+      "learning_rate": 7.771406727828746e-05,
+      "loss": 0.097,
+      "step": 11051
+    },
+    {
+      "epoch": 101.39,
+      "learning_rate": 7.767584097859327e-05,
+      "loss": 0.0974,
+      "step": 11052
+    },
+    {
+      "epoch": 101.4,
+      "learning_rate": 7.763761467889909e-05,
+      "loss": 0.0954,
+      "step": 11053
+    },
+    {
+      "epoch": 101.41,
+      "learning_rate": 7.75993883792049e-05,
+      "loss": 0.0491,
+      "step": 11054
+    },
+    {
+      "epoch": 101.42,
+      "learning_rate": 7.756116207951071e-05,
+      "loss": 0.1046,
+      "step": 11055
+    },
+    {
+      "epoch": 101.43,
+      "learning_rate": 7.752293577981652e-05,
+      "loss": 0.1356,
+      "step": 11056
+    },
+    {
+      "epoch": 101.44,
+      "learning_rate": 7.748470948012234e-05,
+      "loss": 0.1075,
+      "step": 11057
+    },
+    {
+      "epoch": 101.45,
+      "learning_rate": 7.744648318042814e-05,
+      "loss": 0.0698,
+      "step": 11058
+    },
+    {
+      "epoch": 101.46,
+      "learning_rate": 7.740825688073394e-05,
+      "loss": 0.12,
+      "step": 11059
+    },
+    {
+      "epoch": 101.47,
+      "learning_rate": 7.737003058103975e-05,
+      "loss": 0.1658,
+      "step": 11060
+    },
+    {
+      "epoch": 101.48,
+      "learning_rate": 7.733180428134557e-05,
+      "loss": 0.1282,
+      "step": 11061
+    },
+    {
+      "epoch": 101.49,
+      "learning_rate": 7.729357798165137e-05,
+      "loss": 0.179,
+      "step": 11062
+    },
+    {
+      "epoch": 101.5,
+      "learning_rate": 7.725535168195719e-05,
+      "loss": 0.056,
+      "step": 11063
+    },
+    {
+      "epoch": 101.5,
+      "learning_rate": 7.721712538226299e-05,
+      "loss": 0.1875,
+      "step": 11064
+    },
+    {
+      "epoch": 101.51,
+      "learning_rate": 7.717889908256881e-05,
+      "loss": 0.1639,
+      "step": 11065
+    },
+    {
+      "epoch": 101.52,
+      "learning_rate": 7.714067278287463e-05,
+      "loss": 0.1782,
+      "step": 11066
+    },
+    {
+      "epoch": 101.53,
+      "learning_rate": 7.710244648318043e-05,
+      "loss": 0.1536,
+      "step": 11067
+    },
+    {
+      "epoch": 101.54,
+      "learning_rate": 7.706422018348625e-05,
+      "loss": 0.1289,
+      "step": 11068
+    },
+    {
+      "epoch": 101.55,
+      "learning_rate": 7.702599388379205e-05,
+      "loss": 0.0834,
+      "step": 11069
+    },
+    {
+      "epoch": 101.56,
+      "learning_rate": 7.698776758409787e-05,
+      "loss": 0.1472,
+      "step": 11070
+    },
+    {
+      "epoch": 101.57,
+      "learning_rate": 7.694954128440366e-05,
+      "loss": 0.1068,
+      "step": 11071
+    },
+    {
+      "epoch": 101.58,
+      "learning_rate": 7.691131498470948e-05,
+      "loss": 0.1045,
+      "step": 11072
+    },
+    {
+      "epoch": 101.59,
+      "learning_rate": 7.687308868501529e-05,
+      "loss": 0.1141,
+      "step": 11073
+    },
+    {
+      "epoch": 101.6,
+      "learning_rate": 7.68348623853211e-05,
+      "loss": 0.0776,
+      "step": 11074
+    },
+    {
+      "epoch": 101.61,
+      "learning_rate": 7.679663608562691e-05,
+      "loss": 0.1061,
+      "step": 11075
+    },
+    {
+      "epoch": 101.61,
+      "learning_rate": 7.675840978593273e-05,
+      "loss": 0.1435,
+      "step": 11076
+    },
+    {
+      "epoch": 101.62,
+      "learning_rate": 7.672018348623853e-05,
+      "loss": 0.1105,
+      "step": 11077
+    },
+    {
+      "epoch": 101.63,
+      "learning_rate": 7.668195718654435e-05,
+      "loss": 0.1395,
+      "step": 11078
+    },
+    {
+      "epoch": 101.64,
+      "learning_rate": 7.664373088685015e-05,
+      "loss": 0.0876,
+      "step": 11079
+    },
+    {
+      "epoch": 101.65,
+      "learning_rate": 7.660550458715597e-05,
+      "loss": 0.0901,
+      "step": 11080
+    },
+    {
+      "epoch": 101.66,
+      "learning_rate": 7.656727828746179e-05,
+      "loss": 0.1325,
+      "step": 11081
+    },
+    {
+      "epoch": 101.67,
+      "learning_rate": 7.652905198776759e-05,
+      "loss": 0.1022,
+      "step": 11082
+    },
+    {
+      "epoch": 101.68,
+      "learning_rate": 7.64908256880734e-05,
+      "loss": 0.0998,
+      "step": 11083
+    },
+    {
+      "epoch": 101.69,
+      "learning_rate": 7.64525993883792e-05,
+      "loss": 0.1162,
+      "step": 11084
+    },
+    {
+      "epoch": 101.7,
+      "learning_rate": 7.641437308868502e-05,
+      "loss": 0.0981,
+      "step": 11085
+    },
+    {
+      "epoch": 101.71,
+      "learning_rate": 7.637614678899082e-05,
+      "loss": 0.1372,
+      "step": 11086
+    },
+    {
+      "epoch": 101.72,
+      "learning_rate": 7.633792048929664e-05,
+      "loss": 0.1863,
+      "step": 11087
+    },
+    {
+      "epoch": 101.72,
+      "learning_rate": 7.629969418960244e-05,
+      "loss": 0.0779,
+      "step": 11088
+    },
+    {
+      "epoch": 101.73,
+      "learning_rate": 7.626146788990826e-05,
+      "loss": 0.0802,
+      "step": 11089
+    },
+    {
+      "epoch": 101.74,
+      "learning_rate": 7.622324159021407e-05,
+      "loss": 0.1016,
+      "step": 11090
+    },
+    {
+      "epoch": 101.75,
+      "learning_rate": 7.618501529051988e-05,
+      "loss": 0.1495,
+      "step": 11091
+    },
+    {
+      "epoch": 101.76,
+      "learning_rate": 7.614678899082569e-05,
+      "loss": 0.1221,
+      "step": 11092
+    },
+    {
+      "epoch": 101.77,
+      "learning_rate": 7.61085626911315e-05,
+      "loss": 0.1248,
+      "step": 11093
+    },
+    {
+      "epoch": 101.78,
+      "learning_rate": 7.607033639143731e-05,
+      "loss": 0.1786,
+      "step": 11094
+    },
+    {
+      "epoch": 101.79,
+      "learning_rate": 7.603211009174312e-05,
+      "loss": 0.1386,
+      "step": 11095
+    },
+    {
+      "epoch": 101.8,
+      "learning_rate": 7.599388379204893e-05,
+      "loss": 0.1527,
+      "step": 11096
+    },
+    {
+      "epoch": 101.81,
+      "learning_rate": 7.595565749235474e-05,
+      "loss": 0.1038,
+      "step": 11097
+    },
+    {
+      "epoch": 101.82,
+      "learning_rate": 7.591743119266055e-05,
+      "loss": 0.0889,
+      "step": 11098
+    },
+    {
+      "epoch": 101.83,
+      "learning_rate": 7.587920489296636e-05,
+      "loss": 0.1346,
+      "step": 11099
+    },
+    {
+      "epoch": 101.83,
+      "learning_rate": 7.584097859327218e-05,
+      "loss": 0.1581,
+      "step": 11100
+    },
+    {
+      "epoch": 101.84,
+      "learning_rate": 7.580275229357798e-05,
+      "loss": 0.0915,
+      "step": 11101
+    },
+    {
+      "epoch": 101.85,
+      "learning_rate": 7.57645259938838e-05,
+      "loss": 0.0971,
+      "step": 11102
+    },
+    {
+      "epoch": 101.86,
+      "learning_rate": 7.57262996941896e-05,
+      "loss": 0.1084,
+      "step": 11103
+    },
+    {
+      "epoch": 101.87,
+      "learning_rate": 7.568807339449542e-05,
+      "loss": 0.0833,
+      "step": 11104
+    },
+    {
+      "epoch": 101.88,
+      "learning_rate": 7.564984709480123e-05,
+      "loss": 0.1549,
+      "step": 11105
+    },
+    {
+      "epoch": 101.89,
+      "learning_rate": 7.561162079510704e-05,
+      "loss": 0.1318,
+      "step": 11106
+    },
+    {
+      "epoch": 101.9,
+      "learning_rate": 7.557339449541283e-05,
+      "loss": 0.0807,
+      "step": 11107
+    },
+    {
+      "epoch": 101.91,
+      "learning_rate": 7.553516819571865e-05,
+      "loss": 0.0874,
+      "step": 11108
+    },
+    {
+      "epoch": 101.92,
+      "learning_rate": 7.549694189602446e-05,
+      "loss": 0.142,
+      "step": 11109
+    },
+    {
+      "epoch": 101.93,
+      "learning_rate": 7.545871559633027e-05,
+      "loss": 0.1129,
+      "step": 11110
+    },
+    {
+      "epoch": 101.94,
+      "learning_rate": 7.542048929663609e-05,
+      "loss": 0.123,
+      "step": 11111
+    },
+    {
+      "epoch": 101.94,
+      "learning_rate": 7.53822629969419e-05,
+      "loss": 0.1125,
+      "step": 11112
+    },
+    {
+      "epoch": 101.95,
+      "learning_rate": 7.534403669724771e-05,
+      "loss": 0.123,
+      "step": 11113
+    },
+    {
+      "epoch": 101.96,
+      "learning_rate": 7.530581039755352e-05,
+      "loss": 0.0576,
+      "step": 11114
+    },
+    {
+      "epoch": 101.97,
+      "learning_rate": 7.526758409785934e-05,
+      "loss": 0.1386,
+      "step": 11115
+    },
+    {
+      "epoch": 101.98,
+      "learning_rate": 7.522935779816514e-05,
+      "loss": 0.0699,
+      "step": 11116
+    },
+    {
+      "epoch": 101.99,
+      "learning_rate": 7.519113149847096e-05,
+      "loss": 0.1142,
+      "step": 11117
+    },
+    {
+      "epoch": 102.0,
+      "learning_rate": 7.515290519877676e-05,
+      "loss": 0.1734,
+      "step": 11118
+    },
+    {
+      "epoch": 102.01,
+      "learning_rate": 7.511467889908257e-05,
+      "loss": 0.191,
+      "step": 11119
+    },
+    {
+      "epoch": 102.02,
+      "learning_rate": 7.507645259938837e-05,
+      "loss": 0.1476,
+      "step": 11120
+    },
+    {
+      "epoch": 102.03,
+      "learning_rate": 7.503822629969419e-05,
+      "loss": 0.1395,
+      "step": 11121
+    },
+    {
+      "epoch": 102.04,
+      "learning_rate": 7.5e-05,
+      "loss": 0.1528,
+      "step": 11122
+    },
+    {
+      "epoch": 102.05,
+      "learning_rate": 7.496177370030581e-05,
+      "loss": 0.1695,
+      "step": 11123
+    },
+    {
+      "epoch": 102.06,
+      "learning_rate": 7.492354740061162e-05,
+      "loss": 0.0988,
+      "step": 11124
+    },
+    {
+      "epoch": 102.06,
+      "learning_rate": 7.488532110091743e-05,
+      "loss": 0.0951,
+      "step": 11125
+    },
+    {
+      "epoch": 102.07,
+      "learning_rate": 7.484709480122325e-05,
+      "loss": 0.0885,
+      "step": 11126
+    },
+    {
+      "epoch": 102.08,
+      "learning_rate": 7.480886850152906e-05,
+      "loss": 0.0933,
+      "step": 11127
+    },
+    {
+      "epoch": 102.09,
+      "learning_rate": 7.477064220183487e-05,
+      "loss": 0.0872,
+      "step": 11128
+    },
+    {
+      "epoch": 102.1,
+      "learning_rate": 7.473241590214068e-05,
+      "loss": 0.148,
+      "step": 11129
+    },
+    {
+      "epoch": 102.11,
+      "learning_rate": 7.46941896024465e-05,
+      "loss": 0.1187,
+      "step": 11130
+    },
+    {
+      "epoch": 102.12,
+      "learning_rate": 7.465596330275229e-05,
+      "loss": 0.0782,
+      "step": 11131
+    },
+    {
+      "epoch": 102.13,
+      "learning_rate": 7.46177370030581e-05,
+      "loss": 0.096,
+      "step": 11132
+    },
+    {
+      "epoch": 102.14,
+      "learning_rate": 7.457951070336391e-05,
+      "loss": 0.137,
+      "step": 11133
+    },
+    {
+      "epoch": 102.15,
+      "learning_rate": 7.454128440366973e-05,
+      "loss": 0.0954,
+      "step": 11134
+    },
+    {
+      "epoch": 102.16,
+      "learning_rate": 7.450305810397553e-05,
+      "loss": 0.0373,
+      "step": 11135
+    },
+    {
+      "epoch": 102.17,
+      "learning_rate": 7.446483180428135e-05,
+      "loss": 0.1257,
+      "step": 11136
+    },
+    {
+      "epoch": 102.17,
+      "learning_rate": 7.442660550458715e-05,
+      "loss": 0.1016,
+      "step": 11137
+    },
+    {
+      "epoch": 102.18,
+      "learning_rate": 7.438837920489297e-05,
+      "loss": 0.1559,
+      "step": 11138
+    },
+    {
+      "epoch": 102.19,
+      "learning_rate": 7.435015290519877e-05,
+      "loss": 0.0925,
+      "step": 11139
+    },
+    {
+      "epoch": 102.2,
+      "learning_rate": 7.431192660550459e-05,
+      "loss": 0.0864,
+      "step": 11140
+    },
+    {
+      "epoch": 102.21,
+      "learning_rate": 7.427370030581041e-05,
+      "loss": 0.0698,
+      "step": 11141
+    },
+    {
+      "epoch": 102.22,
+      "learning_rate": 7.423547400611621e-05,
+      "loss": 0.1164,
+      "step": 11142
+    },
+    {
+      "epoch": 102.23,
+      "learning_rate": 7.419724770642202e-05,
+      "loss": 0.0591,
+      "step": 11143
+    },
+    {
+      "epoch": 102.24,
+      "learning_rate": 7.415902140672782e-05,
+      "loss": 0.0944,
+      "step": 11144
+    },
+    {
+      "epoch": 102.25,
+      "learning_rate": 7.412079510703364e-05,
+      "loss": 0.0473,
+      "step": 11145
+    },
+    {
+      "epoch": 102.26,
+      "learning_rate": 7.408256880733945e-05,
+      "loss": 0.1818,
+      "step": 11146
+    },
+    {
+      "epoch": 102.27,
+      "learning_rate": 7.404434250764526e-05,
+      "loss": 0.1575,
+      "step": 11147
+    },
+    {
+      "epoch": 102.28,
+      "learning_rate": 7.400611620795107e-05,
+      "loss": 0.1178,
+      "step": 11148
+    },
+    {
+      "epoch": 102.28,
+      "learning_rate": 7.396788990825689e-05,
+      "loss": 0.1755,
+      "step": 11149
+    },
+    {
+      "epoch": 102.29,
+      "learning_rate": 7.392966360856269e-05,
+      "loss": 0.1178,
+      "step": 11150
+    },
+    {
+      "epoch": 102.3,
+      "learning_rate": 7.389143730886851e-05,
+      "loss": 0.1879,
+      "step": 11151
+    },
+    {
+      "epoch": 102.31,
+      "learning_rate": 7.385321100917431e-05,
+      "loss": 0.0894,
+      "step": 11152
+    },
+    {
+      "epoch": 102.32,
+      "learning_rate": 7.381498470948013e-05,
+      "loss": 0.1241,
+      "step": 11153
+    },
+    {
+      "epoch": 102.33,
+      "learning_rate": 7.377675840978593e-05,
+      "loss": 0.0901,
+      "step": 11154
+    },
+    {
+      "epoch": 102.34,
+      "learning_rate": 7.373853211009174e-05,
+      "loss": 0.1234,
+      "step": 11155
+    },
+    {
+      "epoch": 102.35,
+      "learning_rate": 7.370030581039756e-05,
+      "loss": 0.1375,
+      "step": 11156
+    },
+    {
+      "epoch": 102.36,
+      "learning_rate": 7.366207951070336e-05,
+      "loss": 0.1482,
+      "step": 11157
+    },
+    {
+      "epoch": 102.37,
+      "learning_rate": 7.362385321100918e-05,
+      "loss": 0.1418,
+      "step": 11158
+    },
+    {
+      "epoch": 102.38,
+      "learning_rate": 7.358562691131498e-05,
+      "loss": 0.0872,
+      "step": 11159
+    },
+    {
+      "epoch": 102.39,
+      "learning_rate": 7.35474006116208e-05,
+      "loss": 0.0972,
+      "step": 11160
+    },
+    {
+      "epoch": 102.39,
+      "learning_rate": 7.35091743119266e-05,
+      "loss": 0.1231,
+      "step": 11161
+    },
+    {
+      "epoch": 102.4,
+      "learning_rate": 7.347094801223242e-05,
+      "loss": 0.0731,
+      "step": 11162
+    },
+    {
+      "epoch": 102.41,
+      "learning_rate": 7.343272171253823e-05,
+      "loss": 0.1166,
+      "step": 11163
+    },
+    {
+      "epoch": 102.42,
+      "learning_rate": 7.339449541284404e-05,
+      "loss": 0.1103,
+      "step": 11164
+    },
+    {
+      "epoch": 102.43,
+      "learning_rate": 7.335626911314985e-05,
+      "loss": 0.1143,
+      "step": 11165
+    },
+    {
+      "epoch": 102.44,
+      "learning_rate": 7.335626911314985e-05,
+      "loss": 0.1215,
+      "step": 11166
+    },
+    {
+      "epoch": 102.45,
+      "learning_rate": 7.331804281345567e-05,
+      "loss": 0.1698,
+      "step": 11167
+    },
+    {
+      "epoch": 102.46,
+      "learning_rate": 7.327981651376146e-05,
+      "loss": 0.1585,
+      "step": 11168
+    },
+    {
+      "epoch": 102.47,
+      "learning_rate": 7.324159021406728e-05,
+      "loss": 0.0951,
+      "step": 11169
+    },
+    {
+      "epoch": 102.48,
+      "learning_rate": 7.320336391437308e-05,
+      "loss": 0.108,
+      "step": 11170
+    },
+    {
+      "epoch": 102.49,
+      "learning_rate": 7.31651376146789e-05,
+      "loss": 0.0873,
+      "step": 11171
+    },
+    {
+      "epoch": 102.5,
+      "learning_rate": 7.312691131498472e-05,
+      "loss": 0.0659,
+      "step": 11172
+    },
+    {
+      "epoch": 102.5,
+      "learning_rate": 7.308868501529052e-05,
+      "loss": 0.1516,
+      "step": 11173
+    },
+    {
+      "epoch": 102.51,
+      "learning_rate": 7.305045871559634e-05,
+      "loss": 0.1472,
+      "step": 11174
+    },
+    {
+      "epoch": 102.52,
+      "learning_rate": 7.301223241590214e-05,
+      "loss": 0.1883,
+      "step": 11175
+    },
+    {
+      "epoch": 102.53,
+      "learning_rate": 7.297400611620796e-05,
+      "loss": 0.1625,
+      "step": 11176
+    },
+    {
+      "epoch": 102.54,
+      "learning_rate": 7.293577981651376e-05,
+      "loss": 0.1027,
+      "step": 11177
+    },
+    {
+      "epoch": 102.55,
+      "learning_rate": 7.289755351681958e-05,
+      "loss": 0.1776,
+      "step": 11178
+    },
+    {
+      "epoch": 102.56,
+      "learning_rate": 7.285932721712539e-05,
+      "loss": 0.165,
+      "step": 11179
+    },
+    {
+      "epoch": 102.57,
+      "learning_rate": 7.282110091743119e-05,
+      "loss": 0.1194,
+      "step": 11180
+    },
+    {
+      "epoch": 102.58,
+      "learning_rate": 7.2782874617737e-05,
+      "loss": 0.1612,
+      "step": 11181
+    },
+    {
+      "epoch": 102.59,
+      "learning_rate": 7.274464831804281e-05,
+      "loss": 0.1338,
+      "step": 11182
+    },
+    {
+      "epoch": 102.6,
+      "learning_rate": 7.270642201834862e-05,
+      "loss": 0.1754,
+      "step": 11183
+    },
+    {
+      "epoch": 102.61,
+      "learning_rate": 7.266819571865443e-05,
+      "loss": 0.1359,
+      "step": 11184
+    },
+    {
+      "epoch": 102.61,
+      "learning_rate": 7.262996941896025e-05,
+      "loss": 0.1637,
+      "step": 11185
+    },
+    {
+      "epoch": 102.62,
+      "learning_rate": 7.259174311926606e-05,
+      "loss": 0.0867,
+      "step": 11186
+    },
+    {
+      "epoch": 102.63,
+      "learning_rate": 7.255351681957187e-05,
+      "loss": 0.1268,
+      "step": 11187
+    },
+    {
+      "epoch": 102.64,
+      "learning_rate": 7.251529051987768e-05,
+      "loss": 0.1836,
+      "step": 11188
+    },
+    {
+      "epoch": 102.65,
+      "learning_rate": 7.24770642201835e-05,
+      "loss": 0.0632,
+      "step": 11189
+    },
+    {
+      "epoch": 102.66,
+      "learning_rate": 7.24388379204893e-05,
+      "loss": 0.1236,
+      "step": 11190
+    },
+    {
+      "epoch": 102.67,
+      "learning_rate": 7.240061162079512e-05,
+      "loss": 0.1253,
+      "step": 11191
+    },
+    {
+      "epoch": 102.68,
+      "learning_rate": 7.236238532110091e-05,
+      "loss": 0.0994,
+      "step": 11192
+    },
+    {
+      "epoch": 102.69,
+      "learning_rate": 7.232415902140673e-05,
+      "loss": 0.1173,
+      "step": 11193
+    },
+    {
+      "epoch": 102.7,
+      "learning_rate": 7.228593272171253e-05,
+      "loss": 0.1701,
+      "step": 11194
+    },
+    {
+      "epoch": 102.71,
+      "learning_rate": 7.224770642201835e-05,
+      "loss": 0.0897,
+      "step": 11195
+    },
+    {
+      "epoch": 102.72,
+      "learning_rate": 7.220948012232415e-05,
+      "loss": 0.0652,
+      "step": 11196
+    },
+    {
+      "epoch": 102.72,
+      "learning_rate": 7.217125382262997e-05,
+      "loss": 0.1094,
+      "step": 11197
+    },
+    {
+      "epoch": 102.73,
+      "learning_rate": 7.213302752293578e-05,
+      "loss": 0.0742,
+      "step": 11198
+    },
+    {
+      "epoch": 102.74,
+      "learning_rate": 7.20948012232416e-05,
+      "loss": 0.0911,
+      "step": 11199
+    },
+    {
+      "epoch": 102.75,
+      "learning_rate": 7.205657492354741e-05,
+      "loss": 0.1845,
+      "step": 11200
+    },
+    {
+      "epoch": 102.76,
+      "learning_rate": 7.201834862385322e-05,
+      "loss": 0.1504,
+      "step": 11201
+    },
+    {
+      "epoch": 102.77,
+      "learning_rate": 7.198012232415903e-05,
+      "loss": 0.1374,
+      "step": 11202
+    },
+    {
+      "epoch": 102.78,
+      "learning_rate": 7.194189602446484e-05,
+      "loss": 0.0966,
+      "step": 11203
+    },
+    {
+      "epoch": 102.79,
+      "learning_rate": 7.190366972477064e-05,
+      "loss": 0.1248,
+      "step": 11204
+    },
+    {
+      "epoch": 102.8,
+      "learning_rate": 7.186544342507645e-05,
+      "loss": 0.1547,
+      "step": 11205
+    },
+    {
+      "epoch": 102.81,
+      "learning_rate": 7.182721712538226e-05,
+      "loss": 0.1105,
+      "step": 11206
+    },
+    {
+      "epoch": 102.82,
+      "learning_rate": 7.178899082568807e-05,
+      "loss": 0.1084,
+      "step": 11207
+    },
+    {
+      "epoch": 102.83,
+      "learning_rate": 7.175076452599389e-05,
+      "loss": 0.1415,
+      "step": 11208
+    },
+    {
+      "epoch": 102.83,
+      "learning_rate": 7.171253822629969e-05,
+      "loss": 0.0765,
+      "step": 11209
+    },
+    {
+      "epoch": 102.84,
+      "learning_rate": 7.167431192660551e-05,
+      "loss": 0.0887,
+      "step": 11210
+    },
+    {
+      "epoch": 102.85,
+      "learning_rate": 7.163608562691131e-05,
+      "loss": 0.112,
+      "step": 11211
+    },
+    {
+      "epoch": 102.86,
+      "learning_rate": 7.159785932721713e-05,
+      "loss": 0.11,
+      "step": 11212
+    },
+    {
+      "epoch": 102.87,
+      "learning_rate": 7.155963302752293e-05,
+      "loss": 0.123,
+      "step": 11213
+    },
+    {
+      "epoch": 102.88,
+      "learning_rate": 7.152140672782875e-05,
+      "loss": 0.1122,
+      "step": 11214
+    },
+    {
+      "epoch": 102.89,
+      "learning_rate": 7.148318042813457e-05,
+      "loss": 0.1513,
+      "step": 11215
+    },
+    {
+      "epoch": 102.9,
+      "learning_rate": 7.144495412844036e-05,
+      "loss": 0.0862,
+      "step": 11216
+    },
+    {
+      "epoch": 102.91,
+      "learning_rate": 7.140672782874618e-05,
+      "loss": 0.1085,
+      "step": 11217
+    },
+    {
+      "epoch": 102.92,
+      "learning_rate": 7.136850152905198e-05,
+      "loss": 0.2468,
+      "step": 11218
+    },
+    {
+      "epoch": 102.93,
+      "learning_rate": 7.13302752293578e-05,
+      "loss": 0.1524,
+      "step": 11219
+    },
+    {
+      "epoch": 102.94,
+      "learning_rate": 7.12920489296636e-05,
+      "loss": 0.0804,
+      "step": 11220
+    },
+    {
+      "epoch": 102.94,
+      "learning_rate": 7.125382262996942e-05,
+      "loss": 0.0913,
+      "step": 11221
+    },
+    {
+      "epoch": 102.95,
+      "learning_rate": 7.121559633027523e-05,
+      "loss": 0.0671,
+      "step": 11222
+    },
+    {
+      "epoch": 102.96,
+      "learning_rate": 7.117737003058105e-05,
+      "loss": 0.1184,
+      "step": 11223
+    },
+    {
+      "epoch": 102.97,
+      "learning_rate": 7.113914373088685e-05,
+      "loss": 0.163,
+      "step": 11224
+    },
+    {
+      "epoch": 102.98,
+      "learning_rate": 7.110091743119267e-05,
+      "loss": 0.0984,
+      "step": 11225
+    },
+    {
+      "epoch": 102.99,
+      "learning_rate": 7.106269113149847e-05,
+      "loss": 0.0497,
+      "step": 11226
+    },
+    {
+      "epoch": 103.0,
+      "learning_rate": 7.102446483180429e-05,
+      "loss": 0.0899,
+      "step": 11227
+    },
+    {
+      "epoch": 103.01,
+      "learning_rate": 7.098623853211008e-05,
+      "loss": 0.1941,
+      "step": 11228
+    },
+    {
+      "epoch": 103.02,
+      "learning_rate": 7.09480122324159e-05,
+      "loss": 0.1481,
+      "step": 11229
+    },
+    {
+      "epoch": 103.03,
+      "learning_rate": 7.090978593272172e-05,
+      "loss": 0.1091,
+      "step": 11230
+    },
+    {
+      "epoch": 103.04,
+      "learning_rate": 7.087155963302752e-05,
+      "loss": 0.1308,
+      "step": 11231
+    },
+    {
+      "epoch": 103.05,
+      "learning_rate": 7.083333333333334e-05,
+      "loss": 0.1226,
+      "step": 11232
+    },
+    {
+      "epoch": 103.06,
+      "learning_rate": 7.079510703363914e-05,
+      "loss": 0.1157,
+      "step": 11233
+    },
+    {
+      "epoch": 103.06,
+      "learning_rate": 7.075688073394496e-05,
+      "loss": 0.1244,
+      "step": 11234
+    },
+    {
+      "epoch": 103.07,
+      "learning_rate": 7.071865443425076e-05,
+      "loss": 0.1801,
+      "step": 11235
+    },
+    {
+      "epoch": 103.08,
+      "learning_rate": 7.068042813455658e-05,
+      "loss": 0.1436,
+      "step": 11236
+    },
+    {
+      "epoch": 103.09,
+      "learning_rate": 7.064220183486239e-05,
+      "loss": 0.1303,
+      "step": 11237
+    },
+    {
+      "epoch": 103.1,
+      "learning_rate": 7.06039755351682e-05,
+      "loss": 0.0873,
+      "step": 11238
+    },
+    {
+      "epoch": 103.11,
+      "learning_rate": 7.056574923547401e-05,
+      "loss": 0.1155,
+      "step": 11239
+    },
+    {
+      "epoch": 103.12,
+      "learning_rate": 7.052752293577981e-05,
+      "loss": 0.1145,
+      "step": 11240
+    },
+    {
+      "epoch": 103.13,
+      "learning_rate": 7.048929663608562e-05,
+      "loss": 0.1498,
+      "step": 11241
+    },
+    {
+      "epoch": 103.14,
+      "learning_rate": 7.045107033639144e-05,
+      "loss": 0.1176,
+      "step": 11242
+    },
+    {
+      "epoch": 103.15,
+      "learning_rate": 7.041284403669724e-05,
+      "loss": 0.0729,
+      "step": 11243
+    },
+    {
+      "epoch": 103.16,
+      "learning_rate": 7.037461773700306e-05,
+      "loss": 0.0889,
+      "step": 11244
+    },
+    {
+      "epoch": 103.17,
+      "learning_rate": 7.033639143730888e-05,
+      "loss": 0.0649,
+      "step": 11245
+    },
+    {
+      "epoch": 103.17,
+      "learning_rate": 7.029816513761468e-05,
+      "loss": 0.1375,
+      "step": 11246
+    },
+    {
+      "epoch": 103.18,
+      "learning_rate": 7.02599388379205e-05,
+      "loss": 0.1463,
+      "step": 11247
+    },
+    {
+      "epoch": 103.19,
+      "learning_rate": 7.02217125382263e-05,
+      "loss": 0.1142,
+      "step": 11248
+    },
+    {
+      "epoch": 103.2,
+      "learning_rate": 7.018348623853212e-05,
+      "loss": 0.1624,
+      "step": 11249
+    },
+    {
+      "epoch": 103.21,
+      "learning_rate": 7.014525993883792e-05,
+      "loss": 0.0876,
+      "step": 11250
+    },
+    {
+      "epoch": 103.22,
+      "learning_rate": 7.010703363914374e-05,
+      "loss": 0.0545,
+      "step": 11251
+    },
+    {
+      "epoch": 103.23,
+      "learning_rate": 7.006880733944953e-05,
+      "loss": 0.1497,
+      "step": 11252
+    },
+    {
+      "epoch": 103.24,
+      "learning_rate": 7.003058103975535e-05,
+      "loss": 0.1247,
+      "step": 11253
+    },
+    {
+      "epoch": 103.25,
+      "learning_rate": 6.999235474006115e-05,
+      "loss": 0.0988,
+      "step": 11254
+    },
+    {
+      "epoch": 103.26,
+      "learning_rate": 6.995412844036697e-05,
+      "loss": 0.1806,
+      "step": 11255
+    },
+    {
+      "epoch": 103.27,
+      "learning_rate": 6.991590214067278e-05,
+      "loss": 0.1574,
+      "step": 11256
+    },
+    {
+      "epoch": 103.28,
+      "learning_rate": 6.98776758409786e-05,
+      "loss": 0.1737,
+      "step": 11257
+    },
+    {
+      "epoch": 103.28,
+      "learning_rate": 6.98394495412844e-05,
+      "loss": 0.1564,
+      "step": 11258
+    },
+    {
+      "epoch": 103.29,
+      "learning_rate": 6.980122324159022e-05,
+      "loss": 0.1362,
+      "step": 11259
+    },
+    {
+      "epoch": 103.3,
+      "learning_rate": 6.976299694189603e-05,
+      "loss": 0.1037,
+      "step": 11260
+    },
+    {
+      "epoch": 103.31,
+      "learning_rate": 6.972477064220184e-05,
+      "loss": 0.1133,
+      "step": 11261
+    },
+    {
+      "epoch": 103.32,
+      "learning_rate": 6.968654434250766e-05,
+      "loss": 0.1031,
+      "step": 11262
+    },
+    {
+      "epoch": 103.33,
+      "learning_rate": 6.964831804281346e-05,
+      "loss": 0.1548,
+      "step": 11263
+    },
+    {
+      "epoch": 103.34,
+      "learning_rate": 6.961009174311927e-05,
+      "loss": 0.1332,
+      "step": 11264
+    },
+    {
+      "epoch": 103.35,
+      "learning_rate": 6.957186544342507e-05,
+      "loss": 0.1033,
+      "step": 11265
+    },
+    {
+      "epoch": 103.36,
+      "learning_rate": 6.953363914373089e-05,
+      "loss": 0.1327,
+      "step": 11266
+    },
+    {
+      "epoch": 103.37,
+      "learning_rate": 6.949541284403669e-05,
+      "loss": 0.1399,
+      "step": 11267
+    },
+    {
+      "epoch": 103.38,
+      "learning_rate": 6.945718654434251e-05,
+      "loss": 0.1037,
+      "step": 11268
+    },
+    {
+      "epoch": 103.39,
+      "learning_rate": 6.941896024464831e-05,
+      "loss": 0.1038,
+      "step": 11269
+    },
+    {
+      "epoch": 103.39,
+      "learning_rate": 6.938073394495413e-05,
+      "loss": 0.0884,
+      "step": 11270
+    },
+    {
+      "epoch": 103.4,
+      "learning_rate": 6.934250764525994e-05,
+      "loss": 0.1116,
+      "step": 11271
+    },
+    {
+      "epoch": 103.41,
+      "learning_rate": 6.930428134556575e-05,
+      "loss": 0.19,
+      "step": 11272
+    },
+    {
+      "epoch": 103.42,
+      "learning_rate": 6.926605504587156e-05,
+      "loss": 0.1744,
+      "step": 11273
+    },
+    {
+      "epoch": 103.43,
+      "learning_rate": 6.922782874617738e-05,
+      "loss": 0.0599,
+      "step": 11274
+    },
+    {
+      "epoch": 103.44,
+      "learning_rate": 6.91896024464832e-05,
+      "loss": 0.1129,
+      "step": 11275
+    },
+    {
+      "epoch": 103.45,
+      "learning_rate": 6.915137614678898e-05,
+      "loss": 0.125,
+      "step": 11276
+    },
+    {
+      "epoch": 103.46,
+      "learning_rate": 6.91131498470948e-05,
+      "loss": 0.1459,
+      "step": 11277
+    },
+    {
+      "epoch": 103.47,
+      "learning_rate": 6.90749235474006e-05,
+      "loss": 0.0761,
+      "step": 11278
+    },
+    {
+      "epoch": 103.48,
+      "learning_rate": 6.903669724770642e-05,
+      "loss": 0.088,
+      "step": 11279
+    },
+    {
+      "epoch": 103.49,
+      "learning_rate": 6.899847094801223e-05,
+      "loss": 0.0505,
+      "step": 11280
+    },
+    {
+      "epoch": 103.5,
+      "learning_rate": 6.896024464831805e-05,
+      "loss": 0.1263,
+      "step": 11281
+    },
+    {
+      "epoch": 103.5,
+      "learning_rate": 6.892201834862385e-05,
+      "loss": 0.2539,
+      "step": 11282
+    },
+    {
+      "epoch": 103.51,
+      "learning_rate": 6.888379204892967e-05,
+      "loss": 0.1863,
+      "step": 11283
+    },
+    {
+      "epoch": 103.52,
+      "learning_rate": 6.884556574923547e-05,
+      "loss": 0.1314,
+      "step": 11284
+    },
+    {
+      "epoch": 103.53,
+      "learning_rate": 6.880733944954129e-05,
+      "loss": 0.1682,
+      "step": 11285
+    },
+    {
+      "epoch": 103.54,
+      "learning_rate": 6.87691131498471e-05,
+      "loss": 0.1546,
+      "step": 11286
+    },
+    {
+      "epoch": 103.55,
+      "learning_rate": 6.873088685015291e-05,
+      "loss": 0.1144,
+      "step": 11287
+    },
+    {
+      "epoch": 103.56,
+      "learning_rate": 6.86926605504587e-05,
+      "loss": 0.1901,
+      "step": 11288
+    },
+    {
+      "epoch": 103.57,
+      "learning_rate": 6.865443425076452e-05,
+      "loss": 0.1368,
+      "step": 11289
+    },
+    {
+      "epoch": 103.58,
+      "learning_rate": 6.861620795107034e-05,
+      "loss": 0.0873,
+      "step": 11290
+    },
+    {
+      "epoch": 103.59,
+      "learning_rate": 6.857798165137614e-05,
+      "loss": 0.1168,
+      "step": 11291
+    },
+    {
+      "epoch": 103.6,
+      "learning_rate": 6.853975535168196e-05,
+      "loss": 0.1478,
+      "step": 11292
+    },
+    {
+      "epoch": 103.61,
+      "learning_rate": 6.850152905198777e-05,
+      "loss": 0.144,
+      "step": 11293
+    },
+    {
+      "epoch": 103.61,
+      "learning_rate": 6.846330275229358e-05,
+      "loss": 0.1213,
+      "step": 11294
+    },
+    {
+      "epoch": 103.62,
+      "learning_rate": 6.842507645259939e-05,
+      "loss": 0.1511,
+      "step": 11295
+    },
+    {
+      "epoch": 103.63,
+      "learning_rate": 6.83868501529052e-05,
+      "loss": 0.1748,
+      "step": 11296
+    },
+    {
+      "epoch": 103.64,
+      "learning_rate": 6.834862385321101e-05,
+      "loss": 0.1049,
+      "step": 11297
+    },
+    {
+      "epoch": 103.65,
+      "learning_rate": 6.831039755351683e-05,
+      "loss": 0.0577,
+      "step": 11298
+    },
+    {
+      "epoch": 103.66,
+      "learning_rate": 6.827217125382263e-05,
+      "loss": 0.0921,
+      "step": 11299
+    },
+    {
+      "epoch": 103.67,
+      "learning_rate": 6.823394495412844e-05,
+      "loss": 0.0921,
+      "step": 11300
+    },
+    {
+      "epoch": 103.68,
+      "learning_rate": 6.819571865443424e-05,
+      "loss": 0.1061,
+      "step": 11301
+    },
+    {
+      "epoch": 103.69,
+      "learning_rate": 6.815749235474006e-05,
+      "loss": 0.1897,
+      "step": 11302
+    },
+    {
+      "epoch": 103.7,
+      "learning_rate": 6.811926605504588e-05,
+      "loss": 0.1633,
+      "step": 11303
+    },
+    {
+      "epoch": 103.71,
+      "learning_rate": 6.808103975535168e-05,
+      "loss": 0.1004,
+      "step": 11304
+    },
+    {
+      "epoch": 103.72,
+      "learning_rate": 6.80428134556575e-05,
+      "loss": 0.0612,
+      "step": 11305
+    },
+    {
+      "epoch": 103.72,
+      "learning_rate": 6.80045871559633e-05,
+      "loss": 0.2191,
+      "step": 11306
+    },
+    {
+      "epoch": 103.73,
+      "learning_rate": 6.796636085626912e-05,
+      "loss": 0.0994,
+      "step": 11307
+    },
+    {
+      "epoch": 103.74,
+      "learning_rate": 6.792813455657493e-05,
+      "loss": 0.1293,
+      "step": 11308
+    },
+    {
+      "epoch": 103.75,
+      "learning_rate": 6.788990825688074e-05,
+      "loss": 0.1925,
+      "step": 11309
+    },
+    {
+      "epoch": 103.76,
+      "learning_rate": 6.785168195718655e-05,
+      "loss": 0.1621,
+      "step": 11310
+    },
+    {
+      "epoch": 103.77,
+      "learning_rate": 6.781345565749237e-05,
+      "loss": 0.1614,
+      "step": 11311
+    },
+    {
+      "epoch": 103.78,
+      "learning_rate": 6.777522935779816e-05,
+      "loss": 0.1071,
+      "step": 11312
+    },
+    {
+      "epoch": 103.79,
+      "learning_rate": 6.773700305810397e-05,
+      "loss": 0.0989,
+      "step": 11313
+    },
+    {
+      "epoch": 103.8,
+      "learning_rate": 6.769877675840978e-05,
+      "loss": 0.1431,
+      "step": 11314
+    },
+    {
+      "epoch": 103.81,
+      "learning_rate": 6.76605504587156e-05,
+      "loss": 0.1522,
+      "step": 11315
+    },
+    {
+      "epoch": 103.82,
+      "learning_rate": 6.76223241590214e-05,
+      "loss": 0.0904,
+      "step": 11316
+    },
+    {
+      "epoch": 103.83,
+      "learning_rate": 6.758409785932722e-05,
+      "loss": 0.1127,
+      "step": 11317
+    },
+    {
+      "epoch": 103.83,
+      "learning_rate": 6.754587155963304e-05,
+      "loss": 0.11,
+      "step": 11318
+    },
+    {
+      "epoch": 103.84,
+      "learning_rate": 6.750764525993884e-05,
+      "loss": 0.1094,
+      "step": 11319
+    },
+    {
+      "epoch": 103.85,
+      "learning_rate": 6.746941896024466e-05,
+      "loss": 0.0753,
+      "step": 11320
+    },
+    {
+      "epoch": 103.86,
+      "learning_rate": 6.743119266055046e-05,
+      "loss": 0.1576,
+      "step": 11321
+    },
+    {
+      "epoch": 103.87,
+      "learning_rate": 6.739296636085628e-05,
+      "loss": 0.1151,
+      "step": 11322
+    },
+    {
+      "epoch": 103.88,
+      "learning_rate": 6.735474006116208e-05,
+      "loss": 0.0991,
+      "step": 11323
+    },
+    {
+      "epoch": 103.89,
+      "learning_rate": 6.731651376146789e-05,
+      "loss": 0.1016,
+      "step": 11324
+    },
+    {
+      "epoch": 103.9,
+      "learning_rate": 6.727828746177369e-05,
+      "loss": 0.1625,
+      "step": 11325
+    },
+    {
+      "epoch": 103.91,
+      "learning_rate": 6.724006116207951e-05,
+      "loss": 0.0678,
+      "step": 11326
+    },
+    {
+      "epoch": 103.92,
+      "learning_rate": 6.720183486238531e-05,
+      "loss": 0.1259,
+      "step": 11327
+    },
+    {
+      "epoch": 103.93,
+      "learning_rate": 6.716360856269113e-05,
+      "loss": 0.1278,
+      "step": 11328
+    },
+    {
+      "epoch": 103.94,
+      "learning_rate": 6.712538226299694e-05,
+      "loss": 0.1027,
+      "step": 11329
+    },
+    {
+      "epoch": 103.94,
+      "learning_rate": 6.708715596330275e-05,
+      "loss": 0.0429,
+      "step": 11330
+    },
+    {
+      "epoch": 103.95,
+      "learning_rate": 6.704892966360856e-05,
+      "loss": 0.1134,
+      "step": 11331
+    },
+    {
+      "epoch": 103.96,
+      "learning_rate": 6.701070336391438e-05,
+      "loss": 0.1004,
+      "step": 11332
+    },
+    {
+      "epoch": 103.97,
+      "learning_rate": 6.69724770642202e-05,
+      "loss": 0.0996,
+      "step": 11333
+    },
+    {
+      "epoch": 103.98,
+      "learning_rate": 6.6934250764526e-05,
+      "loss": 0.0876,
+      "step": 11334
+    },
+    {
+      "epoch": 103.99,
+      "learning_rate": 6.689602446483182e-05,
+      "loss": 0.0574,
+      "step": 11335
+    },
+    {
+      "epoch": 104.0,
+      "learning_rate": 6.685779816513761e-05,
+      "loss": 0.155,
+      "step": 11336
+    },
+    {
+      "epoch": 104.01,
+      "learning_rate": 6.681957186544343e-05,
+      "loss": 0.2294,
+      "step": 11337
+    },
+    {
+      "epoch": 104.02,
+      "learning_rate": 6.678134556574923e-05,
+      "loss": 0.1138,
+      "step": 11338
+    },
+    {
+      "epoch": 104.03,
+      "learning_rate": 6.674311926605505e-05,
+      "loss": 0.1198,
+      "step": 11339
+    },
+    {
+      "epoch": 104.04,
+      "learning_rate": 6.670489296636085e-05,
+      "loss": 0.1548,
+      "step": 11340
+    },
+    {
+      "epoch": 104.05,
+      "learning_rate": 6.666666666666667e-05,
+      "loss": 0.1431,
+      "step": 11341
+    },
+    {
+      "epoch": 104.06,
+      "learning_rate": 6.662844036697247e-05,
+      "loss": 0.1609,
+      "step": 11342
+    },
+    {
+      "epoch": 104.06,
+      "learning_rate": 6.659021406727829e-05,
+      "loss": 0.1529,
+      "step": 11343
+    },
+    {
+      "epoch": 104.07,
+      "learning_rate": 6.65519877675841e-05,
+      "loss": 0.1122,
+      "step": 11344
+    },
+    {
+      "epoch": 104.08,
+      "learning_rate": 6.651376146788991e-05,
+      "loss": 0.1567,
+      "step": 11345
+    },
+    {
+      "epoch": 104.09,
+      "learning_rate": 6.647553516819572e-05,
+      "loss": 0.0602,
+      "step": 11346
+    },
+    {
+      "epoch": 104.1,
+      "learning_rate": 6.643730886850154e-05,
+      "loss": 0.1439,
+      "step": 11347
+    },
+    {
+      "epoch": 104.11,
+      "learning_rate": 6.639908256880734e-05,
+      "loss": 0.1088,
+      "step": 11348
+    },
+    {
+      "epoch": 104.12,
+      "learning_rate": 6.636085626911314e-05,
+      "loss": 0.1279,
+      "step": 11349
+    },
+    {
+      "epoch": 104.13,
+      "learning_rate": 6.632262996941896e-05,
+      "loss": 0.0929,
+      "step": 11350
+    },
+    {
+      "epoch": 104.14,
+      "learning_rate": 6.628440366972477e-05,
+      "loss": 0.0871,
+      "step": 11351
+    },
+    {
+      "epoch": 104.15,
+      "learning_rate": 6.624617737003058e-05,
+      "loss": 0.0977,
+      "step": 11352
+    },
+    {
+      "epoch": 104.16,
+      "learning_rate": 6.620795107033639e-05,
+      "loss": 0.1222,
+      "step": 11353
+    },
+    {
+      "epoch": 104.17,
+      "learning_rate": 6.616972477064221e-05,
+      "loss": 0.1289,
+      "step": 11354
+    },
+    {
+      "epoch": 104.17,
+      "learning_rate": 6.613149847094801e-05,
+      "loss": 0.1059,
+      "step": 11355
+    },
+    {
+      "epoch": 104.18,
+      "learning_rate": 6.609327217125383e-05,
+      "loss": 0.1447,
+      "step": 11356
+    },
+    {
+      "epoch": 104.19,
+      "learning_rate": 6.605504587155963e-05,
+      "loss": 0.1167,
+      "step": 11357
+    },
+    {
+      "epoch": 104.2,
+      "learning_rate": 6.601681957186545e-05,
+      "loss": 0.1275,
+      "step": 11358
+    },
+    {
+      "epoch": 104.21,
+      "learning_rate": 6.597859327217126e-05,
+      "loss": 0.1227,
+      "step": 11359
+    },
+    {
+      "epoch": 104.22,
+      "learning_rate": 6.594036697247706e-05,
+      "loss": 0.0688,
+      "step": 11360
+    },
+    {
+      "epoch": 104.23,
+      "learning_rate": 6.590214067278286e-05,
+      "loss": 0.0741,
+      "step": 11361
+    },
+    {
+      "epoch": 104.24,
+      "learning_rate": 6.586391437308868e-05,
+      "loss": 0.0573,
+      "step": 11362
+    },
+    {
+      "epoch": 104.25,
+      "learning_rate": 6.58256880733945e-05,
+      "loss": 0.0134,
+      "step": 11363
+    },
+    {
+      "epoch": 104.26,
+      "learning_rate": 6.57874617737003e-05,
+      "loss": 0.1313,
+      "step": 11364
+    },
+    {
+      "epoch": 104.27,
+      "learning_rate": 6.574923547400612e-05,
+      "loss": 0.1145,
+      "step": 11365
+    },
+    {
+      "epoch": 104.28,
+      "learning_rate": 6.571100917431193e-05,
+      "loss": 0.1388,
+      "step": 11366
+    },
+    {
+      "epoch": 104.28,
+      "learning_rate": 6.567278287461774e-05,
+      "loss": 0.1432,
+      "step": 11367
+    },
+    {
+      "epoch": 104.29,
+      "learning_rate": 6.563455657492355e-05,
+      "loss": 0.0947,
+      "step": 11368
+    },
+    {
+      "epoch": 104.3,
+      "learning_rate": 6.559633027522937e-05,
+      "loss": 0.1189,
+      "step": 11369
+    },
+    {
+      "epoch": 104.31,
+      "learning_rate": 6.555810397553517e-05,
+      "loss": 0.0935,
+      "step": 11370
+    },
+    {
+      "epoch": 104.32,
+      "learning_rate": 6.551987767584099e-05,
+      "loss": 0.0994,
+      "step": 11371
+    },
+    {
+      "epoch": 104.33,
+      "learning_rate": 6.548165137614678e-05,
+      "loss": 0.1071,
+      "step": 11372
+    },
+    {
+      "epoch": 104.34,
+      "learning_rate": 6.54434250764526e-05,
+      "loss": 0.1436,
+      "step": 11373
+    },
+    {
+      "epoch": 104.35,
+      "learning_rate": 6.54051987767584e-05,
+      "loss": 0.1256,
+      "step": 11374
+    },
+    {
+      "epoch": 104.36,
+      "learning_rate": 6.536697247706422e-05,
+      "loss": 0.1015,
+      "step": 11375
+    },
+    {
+      "epoch": 104.37,
+      "learning_rate": 6.532874617737002e-05,
+      "loss": 0.1488,
+      "step": 11376
+    },
+    {
+      "epoch": 104.38,
+      "learning_rate": 6.529051987767584e-05,
+      "loss": 0.1438,
+      "step": 11377
+    },
+    {
+      "epoch": 104.39,
+      "learning_rate": 6.525229357798166e-05,
+      "loss": 0.1137,
+      "step": 11378
+    },
+    {
+      "epoch": 104.39,
+      "learning_rate": 6.521406727828746e-05,
+      "loss": 0.1468,
+      "step": 11379
+    },
+    {
+      "epoch": 104.4,
+      "learning_rate": 6.517584097859328e-05,
+      "loss": 0.0868,
+      "step": 11380
+    },
+    {
+      "epoch": 104.41,
+      "learning_rate": 6.513761467889909e-05,
+      "loss": 0.1026,
+      "step": 11381
+    },
+    {
+      "epoch": 104.42,
+      "learning_rate": 6.50993883792049e-05,
+      "loss": 0.1571,
+      "step": 11382
+    },
+    {
+      "epoch": 104.43,
+      "learning_rate": 6.506116207951071e-05,
+      "loss": 0.1149,
+      "step": 11383
+    },
+    {
+      "epoch": 104.44,
+      "learning_rate": 6.502293577981651e-05,
+      "loss": 0.0666,
+      "step": 11384
+    },
+    {
+      "epoch": 104.45,
+      "learning_rate": 6.498470948012232e-05,
+      "loss": 0.1015,
+      "step": 11385
+    },
+    {
+      "epoch": 104.46,
+      "learning_rate": 6.494648318042813e-05,
+      "loss": 0.1462,
+      "step": 11386
+    },
+    {
+      "epoch": 104.47,
+      "learning_rate": 6.490825688073394e-05,
+      "loss": 0.0863,
+      "step": 11387
+    },
+    {
+      "epoch": 104.48,
+      "learning_rate": 6.487003058103976e-05,
+      "loss": 0.0883,
+      "step": 11388
+    },
+    {
+      "epoch": 104.49,
+      "learning_rate": 6.483180428134556e-05,
+      "loss": 0.0725,
+      "step": 11389
+    },
+    {
+      "epoch": 104.5,
+      "learning_rate": 6.479357798165138e-05,
+      "loss": 0.0967,
+      "step": 11390
+    },
+    {
+      "epoch": 104.5,
+      "learning_rate": 6.475535168195718e-05,
+      "loss": 0.196,
+      "step": 11391
+    },
+    {
+      "epoch": 104.51,
+      "learning_rate": 6.4717125382263e-05,
+      "loss": 0.1644,
+      "step": 11392
+    },
+    {
+      "epoch": 104.52,
+      "learning_rate": 6.467889908256882e-05,
+      "loss": 0.1383,
+      "step": 11393
+    },
+    {
+      "epoch": 104.53,
+      "learning_rate": 6.464067278287462e-05,
+      "loss": 0.132,
+      "step": 11394
+    },
+    {
+      "epoch": 104.54,
+      "learning_rate": 6.460244648318044e-05,
+      "loss": 0.1782,
+      "step": 11395
+    },
+    {
+      "epoch": 104.55,
+      "learning_rate": 6.456422018348623e-05,
+      "loss": 0.1295,
+      "step": 11396
+    },
+    {
+      "epoch": 104.56,
+      "learning_rate": 6.452599388379205e-05,
+      "loss": 0.1272,
+      "step": 11397
+    },
+    {
+      "epoch": 104.57,
+      "learning_rate": 6.448776758409785e-05,
+      "loss": 0.0718,
+      "step": 11398
+    },
+    {
+      "epoch": 104.58,
+      "learning_rate": 6.444954128440367e-05,
+      "loss": 0.1174,
+      "step": 11399
+    },
+    {
+      "epoch": 104.59,
+      "learning_rate": 6.441131498470948e-05,
+      "loss": 0.1051,
+      "step": 11400
+    },
+    {
+      "epoch": 104.6,
+      "learning_rate": 6.437308868501529e-05,
+      "loss": 0.134,
+      "step": 11401
+    },
+    {
+      "epoch": 104.61,
+      "learning_rate": 6.43348623853211e-05,
+      "loss": 0.1008,
+      "step": 11402
+    },
+    {
+      "epoch": 104.61,
+      "learning_rate": 6.429663608562692e-05,
+      "loss": 0.0761,
+      "step": 11403
+    },
+    {
+      "epoch": 104.62,
+      "learning_rate": 6.425840978593272e-05,
+      "loss": 0.0731,
+      "step": 11404
+    },
+    {
+      "epoch": 104.63,
+      "learning_rate": 6.422018348623854e-05,
+      "loss": 0.1681,
+      "step": 11405
+    },
+    {
+      "epoch": 104.64,
+      "learning_rate": 6.418195718654436e-05,
+      "loss": 0.0989,
+      "step": 11406
+    },
+    {
+      "epoch": 104.65,
+      "learning_rate": 6.414373088685016e-05,
+      "loss": 0.1082,
+      "step": 11407
+    },
+    {
+      "epoch": 104.66,
+      "learning_rate": 6.410550458715596e-05,
+      "loss": 0.0583,
+      "step": 11408
+    },
+    {
+      "epoch": 104.67,
+      "learning_rate": 6.406727828746177e-05,
+      "loss": 0.0894,
+      "step": 11409
+    },
+    {
+      "epoch": 104.68,
+      "learning_rate": 6.402905198776759e-05,
+      "loss": 0.1302,
+      "step": 11410
+    },
+    {
+      "epoch": 104.69,
+      "learning_rate": 6.399082568807339e-05,
+      "loss": 0.121,
+      "step": 11411
+    },
+    {
+      "epoch": 104.7,
+      "learning_rate": 6.395259938837921e-05,
+      "loss": 0.1465,
+      "step": 11412
+    },
+    {
+      "epoch": 104.71,
+      "learning_rate": 6.391437308868501e-05,
+      "loss": 0.13,
+      "step": 11413
+    },
+    {
+      "epoch": 104.72,
+      "learning_rate": 6.387614678899083e-05,
+      "loss": 0.0753,
+      "step": 11414
+    },
+    {
+      "epoch": 104.72,
+      "learning_rate": 6.383792048929663e-05,
+      "loss": 0.0618,
+      "step": 11415
+    },
+    {
+      "epoch": 104.73,
+      "learning_rate": 6.379969418960245e-05,
+      "loss": 0.0995,
+      "step": 11416
+    },
+    {
+      "epoch": 104.74,
+      "learning_rate": 6.376146788990826e-05,
+      "loss": 0.0745,
+      "step": 11417
+    },
+    {
+      "epoch": 104.75,
+      "learning_rate": 6.372324159021407e-05,
+      "loss": 0.1379,
+      "step": 11418
+    },
+    {
+      "epoch": 104.76,
+      "learning_rate": 6.368501529051988e-05,
+      "loss": 0.1634,
+      "step": 11419
+    },
+    {
+      "epoch": 104.77,
+      "learning_rate": 6.364678899082568e-05,
+      "loss": 0.1276,
+      "step": 11420
+    },
+    {
+      "epoch": 104.78,
+      "learning_rate": 6.36085626911315e-05,
+      "loss": 0.127,
+      "step": 11421
+    },
+    {
+      "epoch": 104.79,
+      "learning_rate": 6.35703363914373e-05,
+      "loss": 0.1125,
+      "step": 11422
+    },
+    {
+      "epoch": 104.8,
+      "learning_rate": 6.353211009174312e-05,
+      "loss": 0.1365,
+      "step": 11423
+    },
+    {
+      "epoch": 104.81,
+      "learning_rate": 6.349388379204893e-05,
+      "loss": 0.0994,
+      "step": 11424
+    },
+    {
+      "epoch": 104.82,
+      "learning_rate": 6.345565749235475e-05,
+      "loss": 0.1503,
+      "step": 11425
+    },
+    {
+      "epoch": 104.83,
+      "learning_rate": 6.341743119266055e-05,
+      "loss": 0.126,
+      "step": 11426
+    },
+    {
+      "epoch": 104.83,
+      "learning_rate": 6.337920489296637e-05,
+      "loss": 0.1108,
+      "step": 11427
+    },
+    {
+      "epoch": 104.84,
+      "learning_rate": 6.334097859327217e-05,
+      "loss": 0.1183,
+      "step": 11428
+    },
+    {
+      "epoch": 104.85,
+      "learning_rate": 6.330275229357799e-05,
+      "loss": 0.0623,
+      "step": 11429
+    },
+    {
+      "epoch": 104.86,
+      "learning_rate": 6.32645259938838e-05,
+      "loss": 0.1117,
+      "step": 11430
+    },
+    {
+      "epoch": 104.87,
+      "learning_rate": 6.322629969418961e-05,
+      "loss": 0.1016,
+      "step": 11431
+    },
+    {
+      "epoch": 104.88,
+      "learning_rate": 6.31880733944954e-05,
+      "loss": 0.0679,
+      "step": 11432
+    },
+    {
+      "epoch": 104.89,
+      "learning_rate": 6.314984709480122e-05,
+      "loss": 0.1478,
+      "step": 11433
+    },
+    {
+      "epoch": 104.9,
+      "learning_rate": 6.311162079510702e-05,
+      "loss": 0.093,
+      "step": 11434
+    },
+    {
+      "epoch": 104.91,
+      "learning_rate": 6.307339449541284e-05,
+      "loss": 0.1683,
+      "step": 11435
+    },
+    {
+      "epoch": 104.92,
+      "learning_rate": 6.303516819571866e-05,
+      "loss": 0.0996,
+      "step": 11436
+    },
+    {
+      "epoch": 104.93,
+      "learning_rate": 6.299694189602446e-05,
+      "loss": 0.1441,
+      "step": 11437
+    },
+    {
+      "epoch": 104.94,
+      "learning_rate": 6.295871559633028e-05,
+      "loss": 0.1693,
+      "step": 11438
+    },
+    {
+      "epoch": 104.94,
+      "learning_rate": 6.292048929663609e-05,
+      "loss": 0.1138,
+      "step": 11439
+    },
+    {
+      "epoch": 104.95,
+      "learning_rate": 6.28822629969419e-05,
+      "loss": 0.1006,
+      "step": 11440
+    },
+    {
+      "epoch": 104.96,
+      "learning_rate": 6.284403669724771e-05,
+      "loss": 0.1205,
+      "step": 11441
+    },
+    {
+      "epoch": 104.97,
+      "learning_rate": 6.280581039755353e-05,
+      "loss": 0.0731,
+      "step": 11442
+    },
+    {
+      "epoch": 104.98,
+      "learning_rate": 6.276758409785933e-05,
+      "loss": 0.1866,
+      "step": 11443
+    },
+    {
+      "epoch": 104.99,
+      "learning_rate": 6.272935779816513e-05,
+      "loss": 0.0313,
+      "step": 11444
+    },
+    {
+      "epoch": 105.0,
+      "learning_rate": 6.269113149847094e-05,
+      "loss": 0.102,
+      "step": 11445
+    },
+    {
+      "epoch": 105.01,
+      "learning_rate": 6.265290519877676e-05,
+      "loss": 0.1632,
+      "step": 11446
+    },
+    {
+      "epoch": 105.02,
+      "learning_rate": 6.261467889908256e-05,
+      "loss": 0.143,
+      "step": 11447
+    },
+    {
+      "epoch": 105.03,
+      "learning_rate": 6.257645259938838e-05,
+      "loss": 0.1529,
+      "step": 11448
+    },
+    {
+      "epoch": 105.04,
+      "learning_rate": 6.253822629969418e-05,
+      "loss": 0.0866,
+      "step": 11449
+    },
+    {
+      "epoch": 105.05,
+      "learning_rate": 6.25e-05,
+      "loss": 0.1579,
+      "step": 11450
+    },
+    {
+      "epoch": 105.06,
+      "learning_rate": 6.246177370030582e-05,
+      "loss": 0.1255,
+      "step": 11451
+    },
+    {
+      "epoch": 105.06,
+      "learning_rate": 6.242354740061162e-05,
+      "loss": 0.1302,
+      "step": 11452
+    },
+    {
+      "epoch": 105.07,
+      "learning_rate": 6.238532110091744e-05,
+      "loss": 0.1074,
+      "step": 11453
+    },
+    {
+      "epoch": 105.08,
+      "learning_rate": 6.234709480122325e-05,
+      "loss": 0.1428,
+      "step": 11454
+    },
+    {
+      "epoch": 105.09,
+      "learning_rate": 6.230886850152905e-05,
+      "loss": 0.0776,
+      "step": 11455
+    },
+    {
+      "epoch": 105.1,
+      "learning_rate": 6.227064220183487e-05,
+      "loss": 0.0935,
+      "step": 11456
+    },
+    {
+      "epoch": 105.11,
+      "learning_rate": 6.223241590214067e-05,
+      "loss": 0.0945,
+      "step": 11457
+    },
+    {
+      "epoch": 105.12,
+      "learning_rate": 6.219418960244649e-05,
+      "loss": 0.1178,
+      "step": 11458
+    },
+    {
+      "epoch": 105.13,
+      "learning_rate": 6.21559633027523e-05,
+      "loss": 0.1224,
+      "step": 11459
+    },
+    {
+      "epoch": 105.14,
+      "learning_rate": 6.21177370030581e-05,
+      "loss": 0.1132,
+      "step": 11460
+    },
+    {
+      "epoch": 105.15,
+      "learning_rate": 6.207951070336392e-05,
+      "loss": 0.1062,
+      "step": 11461
+    },
+    {
+      "epoch": 105.16,
+      "learning_rate": 6.204128440366972e-05,
+      "loss": 0.0491,
+      "step": 11462
+    },
+    {
+      "epoch": 105.17,
+      "learning_rate": 6.200305810397554e-05,
+      "loss": 0.062,
+      "step": 11463
+    },
+    {
+      "epoch": 105.17,
+      "learning_rate": 6.196483180428134e-05,
+      "loss": 0.193,
+      "step": 11464
+    },
+    {
+      "epoch": 105.18,
+      "learning_rate": 6.192660550458716e-05,
+      "loss": 0.1407,
+      "step": 11465
+    },
+    {
+      "epoch": 105.19,
+      "learning_rate": 6.188837920489296e-05,
+      "loss": 0.1149,
+      "step": 11466
+    },
+    {
+      "epoch": 105.2,
+      "learning_rate": 6.185015290519878e-05,
+      "loss": 0.0882,
+      "step": 11467
+    },
+    {
+      "epoch": 105.21,
+      "learning_rate": 6.181192660550459e-05,
+      "loss": 0.1061,
+      "step": 11468
+    },
+    {
+      "epoch": 105.22,
+      "learning_rate": 6.17737003058104e-05,
+      "loss": 0.1255,
+      "step": 11469
+    },
+    {
+      "epoch": 105.23,
+      "learning_rate": 6.173547400611621e-05,
+      "loss": 0.0615,
+      "step": 11470
+    },
+    {
+      "epoch": 105.24,
+      "learning_rate": 6.169724770642203e-05,
+      "loss": 0.0923,
+      "step": 11471
+    },
+    {
+      "epoch": 105.25,
+      "learning_rate": 6.165902140672783e-05,
+      "loss": 0.076,
+      "step": 11472
+    },
+    {
+      "epoch": 105.26,
+      "learning_rate": 6.162079510703364e-05,
+      "loss": 0.1726,
+      "step": 11473
+    },
+    {
+      "epoch": 105.27,
+      "learning_rate": 6.158256880733945e-05,
+      "loss": 0.183,
+      "step": 11474
+    },
+    {
+      "epoch": 105.28,
+      "learning_rate": 6.154434250764526e-05,
+      "loss": 0.1364,
+      "step": 11475
+    },
+    {
+      "epoch": 105.28,
+      "learning_rate": 6.150611620795108e-05,
+      "loss": 0.1449,
+      "step": 11476
+    },
+    {
+      "epoch": 105.29,
+      "learning_rate": 6.146788990825688e-05,
+      "loss": 0.1163,
+      "step": 11477
+    },
+    {
+      "epoch": 105.3,
+      "learning_rate": 6.142966360856268e-05,
+      "loss": 0.1053,
+      "step": 11478
+    },
+    {
+      "epoch": 105.31,
+      "learning_rate": 6.13914373088685e-05,
+      "loss": 0.0769,
+      "step": 11479
+    },
+    {
+      "epoch": 105.32,
+      "learning_rate": 6.135321100917432e-05,
+      "loss": 0.1087,
+      "step": 11480
+    },
+    {
+      "epoch": 105.33,
+      "learning_rate": 6.131498470948012e-05,
+      "loss": 0.1254,
+      "step": 11481
+    },
+    {
+      "epoch": 105.34,
+      "learning_rate": 6.127675840978594e-05,
+      "loss": 0.0864,
+      "step": 11482
+    },
+    {
+      "epoch": 105.35,
+      "learning_rate": 6.123853211009175e-05,
+      "loss": 0.1186,
+      "step": 11483
+    },
+    {
+      "epoch": 105.36,
+      "learning_rate": 6.120030581039755e-05,
+      "loss": 0.1032,
+      "step": 11484
+    },
+    {
+      "epoch": 105.37,
+      "learning_rate": 6.116207951070337e-05,
+      "loss": 0.1424,
+      "step": 11485
+    },
+    {
+      "epoch": 105.38,
+      "learning_rate": 6.112385321100917e-05,
+      "loss": 0.1722,
+      "step": 11486
+    },
+    {
+      "epoch": 105.39,
+      "learning_rate": 6.108562691131499e-05,
+      "loss": 0.1387,
+      "step": 11487
+    },
+    {
+      "epoch": 105.39,
+      "learning_rate": 6.10474006116208e-05,
+      "loss": 0.0699,
+      "step": 11488
+    },
+    {
+      "epoch": 105.4,
+      "learning_rate": 6.100917431192661e-05,
+      "loss": 0.165,
+      "step": 11489
+    },
+    {
+      "epoch": 105.41,
+      "learning_rate": 6.097094801223242e-05,
+      "loss": 0.1588,
+      "step": 11490
+    },
+    {
+      "epoch": 105.42,
+      "learning_rate": 6.093272171253823e-05,
+      "loss": 0.1059,
+      "step": 11491
+    },
+    {
+      "epoch": 105.43,
+      "learning_rate": 6.089449541284404e-05,
+      "loss": 0.1225,
+      "step": 11492
+    },
+    {
+      "epoch": 105.44,
+      "learning_rate": 6.085626911314985e-05,
+      "loss": 0.1148,
+      "step": 11493
+    },
+    {
+      "epoch": 105.45,
+      "learning_rate": 6.081804281345566e-05,
+      "loss": 0.0952,
+      "step": 11494
+    },
+    {
+      "epoch": 105.46,
+      "learning_rate": 6.077981651376147e-05,
+      "loss": 0.0633,
+      "step": 11495
+    },
+    {
+      "epoch": 105.47,
+      "learning_rate": 6.0741590214067276e-05,
+      "loss": 0.0707,
+      "step": 11496
+    },
+    {
+      "epoch": 105.48,
+      "learning_rate": 6.070336391437309e-05,
+      "loss": 0.0671,
+      "step": 11497
+    },
+    {
+      "epoch": 105.49,
+      "learning_rate": 6.06651376146789e-05,
+      "loss": 0.1597,
+      "step": 11498
+    },
+    {
+      "epoch": 105.5,
+      "learning_rate": 6.062691131498471e-05,
+      "loss": 0.0456,
+      "step": 11499
+    },
+    {
+      "epoch": 105.5,
+      "learning_rate": 6.058868501529052e-05,
+      "loss": 0.183,
+      "step": 11500
+    },
+    {
+      "epoch": 105.5,
+      "eval_cer": 0.13916250618450052,
+      "eval_loss": 0.7310239672660828,
+      "eval_runtime": 87.2755,
+      "eval_samples_per_second": 18.871,
+      "eval_steps_per_second": 2.36,
+      "eval_wer": 0.4838116637728526,
+      "step": 11500
+    },
+    {
+      "epoch": 105.51,
+      "learning_rate": 6.055045871559633e-05,
+      "loss": 0.1149,
+      "step": 11501
+    },
+    {
+      "epoch": 105.52,
+      "learning_rate": 6.051223241590214e-05,
+      "loss": 0.1846,
+      "step": 11502
+    },
+    {
+      "epoch": 105.53,
+      "learning_rate": 6.0474006116207954e-05,
+      "loss": 0.1567,
+      "step": 11503
+    },
+    {
+      "epoch": 105.54,
+      "learning_rate": 6.0435779816513765e-05,
+      "loss": 0.1304,
+      "step": 11504
+    },
+    {
+      "epoch": 105.55,
+      "learning_rate": 6.0397553516819576e-05,
+      "loss": 0.1121,
+      "step": 11505
+    },
+    {
+      "epoch": 105.56,
+      "learning_rate": 6.035932721712539e-05,
+      "loss": 0.089,
+      "step": 11506
+    },
+    {
+      "epoch": 105.57,
+      "learning_rate": 6.03211009174312e-05,
+      "loss": 0.1207,
+      "step": 11507
+    },
+    {
+      "epoch": 105.58,
+      "learning_rate": 6.0282874617737e-05,
+      "loss": 0.1284,
+      "step": 11508
+    },
+    {
+      "epoch": 105.59,
+      "learning_rate": 6.0244648318042813e-05,
+      "loss": 0.1378,
+      "step": 11509
+    },
+    {
+      "epoch": 105.6,
+      "learning_rate": 6.0206422018348624e-05,
+      "loss": 0.1494,
+      "step": 11510
+    },
+    {
+      "epoch": 105.61,
+      "learning_rate": 6.0168195718654436e-05,
+      "loss": 0.1252,
+      "step": 11511
+    },
+    {
+      "epoch": 105.61,
+      "learning_rate": 6.012996941896025e-05,
+      "loss": 0.1335,
+      "step": 11512
+    },
+    {
+      "epoch": 105.62,
+      "learning_rate": 6.009174311926606e-05,
+      "loss": 0.0935,
+      "step": 11513
+    },
+    {
+      "epoch": 105.63,
+      "learning_rate": 6.005351681957186e-05,
+      "loss": 0.2125,
+      "step": 11514
+    },
+    {
+      "epoch": 105.64,
+      "learning_rate": 6.001529051987767e-05,
+      "loss": 0.1059,
+      "step": 11515
+    },
+    {
+      "epoch": 105.65,
+      "learning_rate": 5.9977064220183484e-05,
+      "loss": 0.1099,
+      "step": 11516
+    },
+    {
+      "epoch": 105.66,
+      "learning_rate": 5.99388379204893e-05,
+      "loss": 0.1152,
+      "step": 11517
+    },
+    {
+      "epoch": 105.67,
+      "learning_rate": 5.990061162079511e-05,
+      "loss": 0.1102,
+      "step": 11518
+    },
+    {
+      "epoch": 105.68,
+      "learning_rate": 5.9862385321100924e-05,
+      "loss": 0.1047,
+      "step": 11519
+    },
+    {
+      "epoch": 105.69,
+      "learning_rate": 5.982415902140673e-05,
+      "loss": 0.0868,
+      "step": 11520
+    },
+    {
+      "epoch": 105.7,
+      "learning_rate": 5.978593272171254e-05,
+      "loss": 0.1656,
+      "step": 11521
+    },
+    {
+      "epoch": 105.71,
+      "learning_rate": 5.974770642201835e-05,
+      "loss": 0.1092,
+      "step": 11522
+    },
+    {
+      "epoch": 105.72,
+      "learning_rate": 5.970948012232416e-05,
+      "loss": 0.0496,
+      "step": 11523
+    },
+    {
+      "epoch": 105.72,
+      "learning_rate": 5.967125382262997e-05,
+      "loss": 0.084,
+      "step": 11524
+    },
+    {
+      "epoch": 105.73,
+      "learning_rate": 5.9633027522935784e-05,
+      "loss": 0.072,
+      "step": 11525
+    },
+    {
+      "epoch": 105.74,
+      "learning_rate": 5.959480122324159e-05,
+      "loss": 0.0761,
+      "step": 11526
+    },
+    {
+      "epoch": 105.75,
+      "learning_rate": 5.95565749235474e-05,
+      "loss": 0.1955,
+      "step": 11527
+    },
+    {
+      "epoch": 105.76,
+      "learning_rate": 5.951834862385321e-05,
+      "loss": 0.1839,
+      "step": 11528
+    },
+    {
+      "epoch": 105.77,
+      "learning_rate": 5.948012232415902e-05,
+      "loss": 0.1422,
+      "step": 11529
+    },
+    {
+      "epoch": 105.78,
+      "learning_rate": 5.944189602446483e-05,
+      "loss": 0.1684,
+      "step": 11530
+    },
+    {
+      "epoch": 105.79,
+      "learning_rate": 5.940366972477065e-05,
+      "loss": 0.1158,
+      "step": 11531
+    },
+    {
+      "epoch": 105.8,
+      "learning_rate": 5.9365443425076454e-05,
+      "loss": 0.0765,
+      "step": 11532
+    },
+    {
+      "epoch": 105.81,
+      "learning_rate": 5.9327217125382265e-05,
+      "loss": 0.09,
+      "step": 11533
+    },
+    {
+      "epoch": 105.82,
+      "learning_rate": 5.9288990825688076e-05,
+      "loss": 0.1207,
+      "step": 11534
+    },
+    {
+      "epoch": 105.83,
+      "learning_rate": 5.925076452599389e-05,
+      "loss": 0.1188,
+      "step": 11535
+    },
+    {
+      "epoch": 105.83,
+      "learning_rate": 5.92125382262997e-05,
+      "loss": 0.0742,
+      "step": 11536
+    },
+    {
+      "epoch": 105.84,
+      "learning_rate": 5.917431192660551e-05,
+      "loss": 0.0842,
+      "step": 11537
+    },
+    {
+      "epoch": 105.85,
+      "learning_rate": 5.9136085626911314e-05,
+      "loss": 0.1567,
+      "step": 11538
+    },
+    {
+      "epoch": 105.86,
+      "learning_rate": 5.9097859327217125e-05,
+      "loss": 0.1012,
+      "step": 11539
+    },
+    {
+      "epoch": 105.87,
+      "learning_rate": 5.9059633027522936e-05,
+      "loss": 0.1172,
+      "step": 11540
+    },
+    {
+      "epoch": 105.88,
+      "learning_rate": 5.902140672782875e-05,
+      "loss": 0.1014,
+      "step": 11541
+    },
+    {
+      "epoch": 105.89,
+      "learning_rate": 5.898318042813456e-05,
+      "loss": 0.1458,
+      "step": 11542
+    },
+    {
+      "epoch": 105.9,
+      "learning_rate": 5.894495412844037e-05,
+      "loss": 0.1219,
+      "step": 11543
+    },
+    {
+      "epoch": 105.91,
+      "learning_rate": 5.8906727828746174e-05,
+      "loss": 0.0695,
+      "step": 11544
+    },
+    {
+      "epoch": 105.92,
+      "learning_rate": 5.8868501529051985e-05,
+      "loss": 0.1069,
+      "step": 11545
+    },
+    {
+      "epoch": 105.93,
+      "learning_rate": 5.88302752293578e-05,
+      "loss": 0.1152,
+      "step": 11546
+    },
+    {
+      "epoch": 105.94,
+      "learning_rate": 5.8792048929663614e-05,
+      "loss": 0.0983,
+      "step": 11547
+    },
+    {
+      "epoch": 105.94,
+      "learning_rate": 5.8753822629969425e-05,
+      "loss": 0.0939,
+      "step": 11548
+    },
+    {
+      "epoch": 105.95,
+      "learning_rate": 5.8715596330275236e-05,
+      "loss": 0.0931,
+      "step": 11549
+    },
+    {
+      "epoch": 105.96,
+      "learning_rate": 5.867737003058104e-05,
+      "loss": 0.0735,
+      "step": 11550
+    },
+    {
+      "epoch": 105.97,
+      "learning_rate": 5.863914373088685e-05,
+      "loss": 0.0886,
+      "step": 11551
+    },
+    {
+      "epoch": 105.98,
+      "learning_rate": 5.860091743119266e-05,
+      "loss": 0.1104,
+      "step": 11552
+    },
+    {
+      "epoch": 105.99,
+      "learning_rate": 5.856269113149847e-05,
+      "loss": 0.0426,
+      "step": 11553
+    },
+    {
+      "epoch": 106.0,
+      "learning_rate": 5.8524464831804284e-05,
+      "loss": 0.1062,
+      "step": 11554
+    },
+    {
+      "epoch": 106.01,
+      "learning_rate": 5.8486238532110095e-05,
+      "loss": 0.2032,
+      "step": 11555
+    },
+    {
+      "epoch": 106.02,
+      "learning_rate": 5.84480122324159e-05,
+      "loss": 0.1204,
+      "step": 11556
+    },
+    {
+      "epoch": 106.03,
+      "learning_rate": 5.840978593272171e-05,
+      "loss": 0.1138,
+      "step": 11557
+    },
+    {
+      "epoch": 106.04,
+      "learning_rate": 5.837155963302752e-05,
+      "loss": 0.1326,
+      "step": 11558
+    },
+    {
+      "epoch": 106.05,
+      "learning_rate": 5.833333333333333e-05,
+      "loss": 0.1282,
+      "step": 11559
+    },
+    {
+      "epoch": 106.06,
+      "learning_rate": 5.8295107033639144e-05,
+      "loss": 0.1384,
+      "step": 11560
+    },
+    {
+      "epoch": 106.06,
+      "learning_rate": 5.825688073394496e-05,
+      "loss": 0.1203,
+      "step": 11561
+    },
+    {
+      "epoch": 106.07,
+      "learning_rate": 5.8218654434250766e-05,
+      "loss": 0.1184,
+      "step": 11562
+    },
+    {
+      "epoch": 106.08,
+      "learning_rate": 5.818042813455658e-05,
+      "loss": 0.0839,
+      "step": 11563
+    },
+    {
+      "epoch": 106.09,
+      "learning_rate": 5.814220183486239e-05,
+      "loss": 0.1442,
+      "step": 11564
+    },
+    {
+      "epoch": 106.1,
+      "learning_rate": 5.81039755351682e-05,
+      "loss": 0.1065,
+      "step": 11565
+    },
+    {
+      "epoch": 106.11,
+      "learning_rate": 5.806574923547401e-05,
+      "loss": 0.0874,
+      "step": 11566
+    },
+    {
+      "epoch": 106.12,
+      "learning_rate": 5.802752293577982e-05,
+      "loss": 0.1089,
+      "step": 11567
+    },
+    {
+      "epoch": 106.13,
+      "learning_rate": 5.7989296636085626e-05,
+      "loss": 0.1201,
+      "step": 11568
+    },
+    {
+      "epoch": 106.14,
+      "learning_rate": 5.7951070336391437e-05,
+      "loss": 0.0827,
+      "step": 11569
+    },
+    {
+      "epoch": 106.15,
+      "learning_rate": 5.791284403669725e-05,
+      "loss": 0.0763,
+      "step": 11570
+    },
+    {
+      "epoch": 106.16,
+      "learning_rate": 5.787461773700306e-05,
+      "loss": 0.0766,
+      "step": 11571
+    },
+    {
+      "epoch": 106.17,
+      "learning_rate": 5.783639143730887e-05,
+      "loss": 0.094,
+      "step": 11572
+    },
+    {
+      "epoch": 106.17,
+      "learning_rate": 5.779816513761468e-05,
+      "loss": 0.0752,
+      "step": 11573
+    },
+    {
+      "epoch": 106.18,
+      "learning_rate": 5.7759938837920485e-05,
+      "loss": 0.104,
+      "step": 11574
+    },
+    {
+      "epoch": 106.19,
+      "learning_rate": 5.7721712538226296e-05,
+      "loss": 0.0489,
+      "step": 11575
+    },
+    {
+      "epoch": 106.2,
+      "learning_rate": 5.7683486238532114e-05,
+      "loss": 0.1036,
+      "step": 11576
+    },
+    {
+      "epoch": 106.21,
+      "learning_rate": 5.7645259938837925e-05,
+      "loss": 0.0757,
+      "step": 11577
+    },
+    {
+      "epoch": 106.22,
+      "learning_rate": 5.7607033639143736e-05,
+      "loss": 0.0562,
+      "step": 11578
+    },
+    {
+      "epoch": 106.23,
+      "learning_rate": 5.756880733944955e-05,
+      "loss": 0.0309,
+      "step": 11579
+    },
+    {
+      "epoch": 106.24,
+      "learning_rate": 5.753058103975535e-05,
+      "loss": 0.0858,
+      "step": 11580
+    },
+    {
+      "epoch": 106.25,
+      "learning_rate": 5.749235474006116e-05,
+      "loss": 0.0731,
+      "step": 11581
+    },
+    {
+      "epoch": 106.26,
+      "learning_rate": 5.7454128440366974e-05,
+      "loss": 0.1881,
+      "step": 11582
+    },
+    {
+      "epoch": 106.27,
+      "learning_rate": 5.7415902140672785e-05,
+      "loss": 0.207,
+      "step": 11583
+    },
+    {
+      "epoch": 106.28,
+      "learning_rate": 5.7377675840978596e-05,
+      "loss": 0.1399,
+      "step": 11584
+    },
+    {
+      "epoch": 106.28,
+      "learning_rate": 5.733944954128441e-05,
+      "loss": 0.1141,
+      "step": 11585
+    },
+    {
+      "epoch": 106.29,
+      "learning_rate": 5.730122324159021e-05,
+      "loss": 0.2015,
+      "step": 11586
+    },
+    {
+      "epoch": 106.3,
+      "learning_rate": 5.726299694189602e-05,
+      "loss": 0.0885,
+      "step": 11587
+    },
+    {
+      "epoch": 106.31,
+      "learning_rate": 5.722477064220183e-05,
+      "loss": 0.0874,
+      "step": 11588
+    },
+    {
+      "epoch": 106.32,
+      "learning_rate": 5.7186544342507644e-05,
+      "loss": 0.1123,
+      "step": 11589
+    },
+    {
+      "epoch": 106.33,
+      "learning_rate": 5.714831804281346e-05,
+      "loss": 0.1209,
+      "step": 11590
+    },
+    {
+      "epoch": 106.34,
+      "learning_rate": 5.711009174311927e-05,
+      "loss": 0.1151,
+      "step": 11591
+    },
+    {
+      "epoch": 106.35,
+      "learning_rate": 5.707186544342508e-05,
+      "loss": 0.0776,
+      "step": 11592
+    },
+    {
+      "epoch": 106.36,
+      "learning_rate": 5.703363914373089e-05,
+      "loss": 0.0875,
+      "step": 11593
+    },
+    {
+      "epoch": 106.37,
+      "learning_rate": 5.69954128440367e-05,
+      "loss": 0.0764,
+      "step": 11594
+    },
+    {
+      "epoch": 106.38,
+      "learning_rate": 5.695718654434251e-05,
+      "loss": 0.1086,
+      "step": 11595
+    },
+    {
+      "epoch": 106.39,
+      "learning_rate": 5.691896024464832e-05,
+      "loss": 0.097,
+      "step": 11596
+    },
+    {
+      "epoch": 106.39,
+      "learning_rate": 5.688073394495413e-05,
+      "loss": 0.11,
+      "step": 11597
+    },
+    {
+      "epoch": 106.4,
+      "learning_rate": 5.684250764525994e-05,
+      "loss": 0.1089,
+      "step": 11598
+    },
+    {
+      "epoch": 106.41,
+      "learning_rate": 5.680428134556575e-05,
+      "loss": 0.0564,
+      "step": 11599
+    },
+    {
+      "epoch": 106.42,
+      "learning_rate": 5.676605504587156e-05,
+      "loss": 0.1189,
+      "step": 11600
+    },
+    {
+      "epoch": 106.43,
+      "learning_rate": 5.672782874617737e-05,
+      "loss": 0.1014,
+      "step": 11601
+    },
+    {
+      "epoch": 106.44,
+      "learning_rate": 5.668960244648318e-05,
+      "loss": 0.1162,
+      "step": 11602
+    },
+    {
+      "epoch": 106.45,
+      "learning_rate": 5.665137614678899e-05,
+      "loss": 0.1201,
+      "step": 11603
+    },
+    {
+      "epoch": 106.46,
+      "learning_rate": 5.66131498470948e-05,
+      "loss": 0.1046,
+      "step": 11604
+    },
+    {
+      "epoch": 106.47,
+      "learning_rate": 5.6574923547400615e-05,
+      "loss": 0.1008,
+      "step": 11605
+    },
+    {
+      "epoch": 106.48,
+      "learning_rate": 5.6536697247706426e-05,
+      "loss": 0.0986,
+      "step": 11606
+    },
+    {
+      "epoch": 106.49,
+      "learning_rate": 5.649847094801224e-05,
+      "loss": 0.0808,
+      "step": 11607
+    },
+    {
+      "epoch": 106.5,
+      "learning_rate": 5.646024464831805e-05,
+      "loss": 0.1069,
+      "step": 11608
+    },
+    {
+      "epoch": 106.5,
+      "learning_rate": 5.642201834862386e-05,
+      "loss": 0.1777,
+      "step": 11609
+    },
+    {
+      "epoch": 106.51,
+      "learning_rate": 5.638379204892966e-05,
+      "loss": 0.1382,
+      "step": 11610
+    },
+    {
+      "epoch": 106.52,
+      "learning_rate": 5.6345565749235474e-05,
+      "loss": 0.1161,
+      "step": 11611
+    },
+    {
+      "epoch": 106.53,
+      "learning_rate": 5.6307339449541285e-05,
+      "loss": 0.1498,
+      "step": 11612
+    },
+    {
+      "epoch": 106.54,
+      "learning_rate": 5.6269113149847096e-05,
+      "loss": 0.0779,
+      "step": 11613
+    },
+    {
+      "epoch": 106.55,
+      "learning_rate": 5.623088685015291e-05,
+      "loss": 0.1076,
+      "step": 11614
+    },
+    {
+      "epoch": 106.56,
+      "learning_rate": 5.619266055045872e-05,
+      "loss": 0.0796,
+      "step": 11615
+    },
+    {
+      "epoch": 106.57,
+      "learning_rate": 5.615443425076452e-05,
+      "loss": 0.1182,
+      "step": 11616
+    },
+    {
+      "epoch": 106.58,
+      "learning_rate": 5.6116207951070334e-05,
+      "loss": 0.081,
+      "step": 11617
+    },
+    {
+      "epoch": 106.59,
+      "learning_rate": 5.6077981651376145e-05,
+      "loss": 0.1399,
+      "step": 11618
+    },
+    {
+      "epoch": 106.6,
+      "learning_rate": 5.6039755351681956e-05,
+      "loss": 0.1031,
+      "step": 11619
+    },
+    {
+      "epoch": 106.61,
+      "learning_rate": 5.6001529051987774e-05,
+      "loss": 0.1391,
+      "step": 11620
+    },
+    {
+      "epoch": 106.61,
+      "learning_rate": 5.5963302752293585e-05,
+      "loss": 0.0922,
+      "step": 11621
+    },
+    {
+      "epoch": 106.62,
+      "learning_rate": 5.592507645259939e-05,
+      "loss": 0.1174,
+      "step": 11622
+    },
+    {
+      "epoch": 106.63,
+      "learning_rate": 5.58868501529052e-05,
+      "loss": 0.0506,
+      "step": 11623
+    },
+    {
+      "epoch": 106.64,
+      "learning_rate": 5.584862385321101e-05,
+      "loss": 0.1023,
+      "step": 11624
+    },
+    {
+      "epoch": 106.65,
+      "learning_rate": 5.581039755351682e-05,
+      "loss": 0.0876,
+      "step": 11625
+    },
+    {
+      "epoch": 106.66,
+      "learning_rate": 5.577217125382263e-05,
+      "loss": 0.0809,
+      "step": 11626
+    },
+    {
+      "epoch": 106.67,
+      "learning_rate": 5.5733944954128444e-05,
+      "loss": 0.1515,
+      "step": 11627
+    },
+    {
+      "epoch": 106.68,
+      "learning_rate": 5.569571865443425e-05,
+      "loss": 0.1335,
+      "step": 11628
+    },
+    {
+      "epoch": 106.69,
+      "learning_rate": 5.565749235474006e-05,
+      "loss": 0.1391,
+      "step": 11629
+    },
+    {
+      "epoch": 106.7,
+      "learning_rate": 5.561926605504587e-05,
+      "loss": 0.0829,
+      "step": 11630
+    },
+    {
+      "epoch": 106.71,
+      "learning_rate": 5.558103975535168e-05,
+      "loss": 0.0618,
+      "step": 11631
+    },
+    {
+      "epoch": 106.72,
+      "learning_rate": 5.554281345565749e-05,
+      "loss": 0.09,
+      "step": 11632
+    },
+    {
+      "epoch": 106.72,
+      "learning_rate": 5.5504587155963304e-05,
+      "loss": 0.104,
+      "step": 11633
+    },
+    {
+      "epoch": 106.73,
+      "learning_rate": 5.546636085626911e-05,
+      "loss": 0.0635,
+      "step": 11634
+    },
+    {
+      "epoch": 106.74,
+      "learning_rate": 5.5428134556574926e-05,
+      "loss": 0.0582,
+      "step": 11635
+    },
+    {
+      "epoch": 106.75,
+      "learning_rate": 5.538990825688074e-05,
+      "loss": 0.1867,
+      "step": 11636
+    },
+    {
+      "epoch": 106.76,
+      "learning_rate": 5.535168195718655e-05,
+      "loss": 0.1426,
+      "step": 11637
+    },
+    {
+      "epoch": 106.77,
+      "learning_rate": 5.531345565749236e-05,
+      "loss": 0.1135,
+      "step": 11638
+    },
+    {
+      "epoch": 106.78,
+      "learning_rate": 5.527522935779817e-05,
+      "loss": 0.0902,
+      "step": 11639
+    },
+    {
+      "epoch": 106.79,
+      "learning_rate": 5.5237003058103975e-05,
+      "loss": 0.0947,
+      "step": 11640
+    },
+    {
+      "epoch": 106.8,
+      "learning_rate": 5.5198776758409786e-05,
+      "loss": 0.1,
+      "step": 11641
+    },
+    {
+      "epoch": 106.81,
+      "learning_rate": 5.51605504587156e-05,
+      "loss": 0.1338,
+      "step": 11642
+    },
+    {
+      "epoch": 106.82,
+      "learning_rate": 5.512232415902141e-05,
+      "loss": 0.093,
+      "step": 11643
+    },
+    {
+      "epoch": 106.83,
+      "learning_rate": 5.508409785932722e-05,
+      "loss": 0.0862,
+      "step": 11644
+    },
+    {
+      "epoch": 106.83,
+      "learning_rate": 5.504587155963303e-05,
+      "loss": 0.1616,
+      "step": 11645
+    },
+    {
+      "epoch": 106.84,
+      "learning_rate": 5.5007645259938834e-05,
+      "loss": 0.1099,
+      "step": 11646
+    },
+    {
+      "epoch": 106.85,
+      "learning_rate": 5.4969418960244645e-05,
+      "loss": 0.0975,
+      "step": 11647
+    },
+    {
+      "epoch": 106.86,
+      "learning_rate": 5.4931192660550456e-05,
+      "loss": 0.1055,
+      "step": 11648
+    },
+    {
+      "epoch": 106.87,
+      "learning_rate": 5.4892966360856274e-05,
+      "loss": 0.1799,
+      "step": 11649
+    },
+    {
+      "epoch": 106.88,
+      "learning_rate": 5.4854740061162085e-05,
+      "loss": 0.1202,
+      "step": 11650
+    },
+    {
+      "epoch": 106.89,
+      "learning_rate": 5.4816513761467896e-05,
+      "loss": 0.1298,
+      "step": 11651
+    },
+    {
+      "epoch": 106.9,
+      "learning_rate": 5.47782874617737e-05,
+      "loss": 0.0825,
+      "step": 11652
+    },
+    {
+      "epoch": 106.91,
+      "learning_rate": 5.474006116207951e-05,
+      "loss": 0.0928,
+      "step": 11653
+    },
+    {
+      "epoch": 106.92,
+      "learning_rate": 5.470183486238532e-05,
+      "loss": 0.0623,
+      "step": 11654
+    },
+    {
+      "epoch": 106.93,
+      "learning_rate": 5.4663608562691134e-05,
+      "loss": 0.1421,
+      "step": 11655
+    },
+    {
+      "epoch": 106.94,
+      "learning_rate": 5.4625382262996945e-05,
+      "loss": 0.0974,
+      "step": 11656
+    },
+    {
+      "epoch": 106.94,
+      "learning_rate": 5.4587155963302756e-05,
+      "loss": 0.1427,
+      "step": 11657
+    },
+    {
+      "epoch": 106.95,
+      "learning_rate": 5.454892966360856e-05,
+      "loss": 0.1166,
+      "step": 11658
+    },
+    {
+      "epoch": 106.96,
+      "learning_rate": 5.451070336391437e-05,
+      "loss": 0.1276,
+      "step": 11659
+    },
+    {
+      "epoch": 106.97,
+      "learning_rate": 5.447247706422018e-05,
+      "loss": 0.0465,
+      "step": 11660
+    },
+    {
+      "epoch": 106.98,
+      "learning_rate": 5.4434250764525994e-05,
+      "loss": 0.0645,
+      "step": 11661
+    },
+    {
+      "epoch": 106.99,
+      "learning_rate": 5.4396024464831805e-05,
+      "loss": 0.0682,
+      "step": 11662
+    },
+    {
+      "epoch": 107.0,
+      "learning_rate": 5.4357798165137616e-05,
+      "loss": 0.1218,
+      "step": 11663
+    },
+    {
+      "epoch": 107.01,
+      "learning_rate": 5.431957186544343e-05,
+      "loss": 0.1622,
+      "step": 11664
+    },
+    {
+      "epoch": 107.02,
+      "learning_rate": 5.428134556574924e-05,
+      "loss": 0.1284,
+      "step": 11665
+    },
+    {
+      "epoch": 107.03,
+      "learning_rate": 5.424311926605505e-05,
+      "loss": 0.1085,
+      "step": 11666
+    },
+    {
+      "epoch": 107.04,
+      "learning_rate": 5.420489296636086e-05,
+      "loss": 0.0532,
+      "step": 11667
+    },
+    {
+      "epoch": 107.05,
+      "learning_rate": 5.416666666666667e-05,
+      "loss": 0.149,
+      "step": 11668
+    },
+    {
+      "epoch": 107.06,
+      "learning_rate": 5.412844036697248e-05,
+      "loss": 0.1032,
+      "step": 11669
+    },
+    {
+      "epoch": 107.06,
+      "learning_rate": 5.4090214067278286e-05,
+      "loss": 0.1112,
+      "step": 11670
+    },
+    {
+      "epoch": 107.07,
+      "learning_rate": 5.40519877675841e-05,
+      "loss": 0.0755,
+      "step": 11671
+    },
+    {
+      "epoch": 107.08,
+      "learning_rate": 5.401376146788991e-05,
+      "loss": 0.0646,
+      "step": 11672
+    },
+    {
+      "epoch": 107.09,
+      "learning_rate": 5.397553516819572e-05,
+      "loss": 0.091,
+      "step": 11673
+    },
+    {
+      "epoch": 107.1,
+      "learning_rate": 5.393730886850153e-05,
+      "loss": 0.1021,
+      "step": 11674
+    },
+    {
+      "epoch": 107.11,
+      "learning_rate": 5.389908256880734e-05,
+      "loss": 0.1521,
+      "step": 11675
+    },
+    {
+      "epoch": 107.12,
+      "learning_rate": 5.3860856269113146e-05,
+      "loss": 0.1017,
+      "step": 11676
+    },
+    {
+      "epoch": 107.13,
+      "learning_rate": 5.382262996941896e-05,
+      "loss": 0.108,
+      "step": 11677
+    },
+    {
+      "epoch": 107.14,
+      "learning_rate": 5.378440366972477e-05,
+      "loss": 0.1644,
+      "step": 11678
+    },
+    {
+      "epoch": 107.15,
+      "learning_rate": 5.3746177370030586e-05,
+      "loss": 0.0953,
+      "step": 11679
+    },
+    {
+      "epoch": 107.16,
+      "learning_rate": 5.37079510703364e-05,
+      "loss": 0.0849,
+      "step": 11680
+    },
+    {
+      "epoch": 107.17,
+      "learning_rate": 5.366972477064221e-05,
+      "loss": 0.132,
+      "step": 11681
+    },
+    {
+      "epoch": 107.17,
+      "learning_rate": 5.363149847094801e-05,
+      "loss": 0.0911,
+      "step": 11682
+    },
+    {
+      "epoch": 107.18,
+      "learning_rate": 5.359327217125382e-05,
+      "loss": 0.1281,
+      "step": 11683
+    },
+    {
+      "epoch": 107.19,
+      "learning_rate": 5.3555045871559634e-05,
+      "loss": 0.0747,
+      "step": 11684
+    },
+    {
+      "epoch": 107.2,
+      "learning_rate": 5.3516819571865445e-05,
+      "loss": 0.0964,
+      "step": 11685
+    },
+    {
+      "epoch": 107.21,
+      "learning_rate": 5.3478593272171257e-05,
+      "loss": 0.094,
+      "step": 11686
+    },
+    {
+      "epoch": 107.22,
+      "learning_rate": 5.344036697247707e-05,
+      "loss": 0.1391,
+      "step": 11687
+    },
+    {
+      "epoch": 107.23,
+      "learning_rate": 5.340214067278287e-05,
+      "loss": 0.0954,
+      "step": 11688
+    },
+    {
+      "epoch": 107.24,
+      "learning_rate": 5.336391437308868e-05,
+      "loss": 0.0806,
+      "step": 11689
+    },
+    {
+      "epoch": 107.25,
+      "learning_rate": 5.3325688073394494e-05,
+      "loss": 0.0536,
+      "step": 11690
+    },
+    {
+      "epoch": 107.26,
+      "learning_rate": 5.3287461773700305e-05,
+      "loss": 0.1382,
+      "step": 11691
+    },
+    {
+      "epoch": 107.27,
+      "learning_rate": 5.3249235474006116e-05,
+      "loss": 0.1509,
+      "step": 11692
+    },
+    {
+      "epoch": 107.28,
+      "learning_rate": 5.3211009174311934e-05,
+      "loss": 0.1001,
+      "step": 11693
+    },
+    {
+      "epoch": 107.28,
+      "learning_rate": 5.317278287461774e-05,
+      "loss": 0.1319,
+      "step": 11694
+    },
+    {
+      "epoch": 107.29,
+      "learning_rate": 5.313455657492355e-05,
+      "loss": 0.1095,
+      "step": 11695
+    },
+    {
+      "epoch": 107.3,
+      "learning_rate": 5.309633027522936e-05,
+      "loss": 0.0815,
+      "step": 11696
+    },
+    {
+      "epoch": 107.31,
+      "learning_rate": 5.305810397553517e-05,
+      "loss": 0.1518,
+      "step": 11697
+    },
+    {
+      "epoch": 107.32,
+      "learning_rate": 5.301987767584098e-05,
+      "loss": 0.1402,
+      "step": 11698
+    },
+    {
+      "epoch": 107.33,
+      "learning_rate": 5.2981651376146794e-05,
+      "loss": 0.1094,
+      "step": 11699
+    },
+    {
+      "epoch": 107.34,
+      "learning_rate": 5.29434250764526e-05,
+      "loss": 0.0687,
+      "step": 11700
+    },
+    {
+      "epoch": 107.35,
+      "learning_rate": 5.290519877675841e-05,
+      "loss": 0.1222,
+      "step": 11701
+    },
+    {
+      "epoch": 107.36,
+      "learning_rate": 5.286697247706422e-05,
+      "loss": 0.0862,
+      "step": 11702
+    },
+    {
+      "epoch": 107.37,
+      "learning_rate": 5.282874617737003e-05,
+      "loss": 0.0944,
+      "step": 11703
+    },
+    {
+      "epoch": 107.38,
+      "learning_rate": 5.279051987767584e-05,
+      "loss": 0.053,
+      "step": 11704
+    },
+    {
+      "epoch": 107.39,
+      "learning_rate": 5.275229357798165e-05,
+      "loss": 0.0615,
+      "step": 11705
+    },
+    {
+      "epoch": 107.39,
+      "learning_rate": 5.271406727828746e-05,
+      "loss": 0.1179,
+      "step": 11706
+    },
+    {
+      "epoch": 107.4,
+      "learning_rate": 5.267584097859327e-05,
+      "loss": 0.1071,
+      "step": 11707
+    },
+    {
+      "epoch": 107.41,
+      "learning_rate": 5.2637614678899086e-05,
+      "loss": 0.1362,
+      "step": 11708
+    },
+    {
+      "epoch": 107.42,
+      "learning_rate": 5.25993883792049e-05,
+      "loss": 0.0738,
+      "step": 11709
+    },
+    {
+      "epoch": 107.43,
+      "learning_rate": 5.256116207951071e-05,
+      "loss": 0.0996,
+      "step": 11710
+    },
+    {
+      "epoch": 107.44,
+      "learning_rate": 5.252293577981652e-05,
+      "loss": 0.0708,
+      "step": 11711
+    },
+    {
+      "epoch": 107.45,
+      "learning_rate": 5.2484709480122324e-05,
+      "loss": 0.1078,
+      "step": 11712
+    },
+    {
+      "epoch": 107.46,
+      "learning_rate": 5.2446483180428135e-05,
+      "loss": 0.0393,
+      "step": 11713
+    },
+    {
+      "epoch": 107.47,
+      "learning_rate": 5.2408256880733946e-05,
+      "loss": 0.1037,
+      "step": 11714
+    },
+    {
+      "epoch": 107.48,
+      "learning_rate": 5.237003058103976e-05,
+      "loss": 0.1404,
+      "step": 11715
+    },
+    {
+      "epoch": 107.49,
+      "learning_rate": 5.233180428134557e-05,
+      "loss": 0.1171,
+      "step": 11716
+    },
+    {
+      "epoch": 107.5,
+      "learning_rate": 5.229357798165138e-05,
+      "loss": 0.0624,
+      "step": 11717
+    },
+    {
+      "epoch": 107.5,
+      "learning_rate": 5.2255351681957184e-05,
+      "loss": 0.1997,
+      "step": 11718
+    },
+    {
+      "epoch": 107.51,
+      "learning_rate": 5.2217125382262995e-05,
+      "loss": 0.1714,
+      "step": 11719
+    },
+    {
+      "epoch": 107.52,
+      "learning_rate": 5.2178899082568806e-05,
+      "loss": 0.1157,
+      "step": 11720
+    },
+    {
+      "epoch": 107.53,
+      "learning_rate": 5.214067278287462e-05,
+      "loss": 0.1192,
+      "step": 11721
+    },
+    {
+      "epoch": 107.54,
+      "learning_rate": 5.210244648318043e-05,
+      "loss": 0.0757,
+      "step": 11722
+    },
+    {
+      "epoch": 107.55,
+      "learning_rate": 5.2064220183486246e-05,
+      "loss": 0.1128,
+      "step": 11723
+    },
+    {
+      "epoch": 107.56,
+      "learning_rate": 5.202599388379205e-05,
+      "loss": 0.0894,
+      "step": 11724
+    },
+    {
+      "epoch": 107.57,
+      "learning_rate": 5.198776758409786e-05,
+      "loss": 0.2193,
+      "step": 11725
+    },
+    {
+      "epoch": 107.58,
+      "learning_rate": 5.194954128440367e-05,
+      "loss": 0.1408,
+      "step": 11726
+    },
+    {
+      "epoch": 107.59,
+      "learning_rate": 5.191131498470948e-05,
+      "loss": 0.1915,
+      "step": 11727
+    },
+    {
+      "epoch": 107.6,
+      "learning_rate": 5.1873088685015294e-05,
+      "loss": 0.1049,
+      "step": 11728
+    },
+    {
+      "epoch": 107.61,
+      "learning_rate": 5.1834862385321105e-05,
+      "loss": 0.0807,
+      "step": 11729
+    },
+    {
+      "epoch": 107.61,
+      "learning_rate": 5.179663608562691e-05,
+      "loss": 0.1106,
+      "step": 11730
+    },
+    {
+      "epoch": 107.62,
+      "learning_rate": 5.175840978593272e-05,
+      "loss": 0.1258,
+      "step": 11731
+    },
+    {
+      "epoch": 107.63,
+      "learning_rate": 5.172018348623853e-05,
+      "loss": 0.101,
+      "step": 11732
+    },
+    {
+      "epoch": 107.64,
+      "learning_rate": 5.168195718654434e-05,
+      "loss": 0.0994,
+      "step": 11733
+    },
+    {
+      "epoch": 107.65,
+      "learning_rate": 5.1643730886850154e-05,
+      "loss": 0.1093,
+      "step": 11734
+    },
+    {
+      "epoch": 107.66,
+      "learning_rate": 5.1605504587155965e-05,
+      "loss": 0.0916,
+      "step": 11735
+    },
+    {
+      "epoch": 107.67,
+      "learning_rate": 5.156727828746177e-05,
+      "loss": 0.105,
+      "step": 11736
+    },
+    {
+      "epoch": 107.68,
+      "learning_rate": 5.152905198776758e-05,
+      "loss": 0.0835,
+      "step": 11737
+    },
+    {
+      "epoch": 107.69,
+      "learning_rate": 5.14908256880734e-05,
+      "loss": 0.0863,
+      "step": 11738
+    },
+    {
+      "epoch": 107.7,
+      "learning_rate": 5.145259938837921e-05,
+      "loss": 0.1619,
+      "step": 11739
+    },
+    {
+      "epoch": 107.71,
+      "learning_rate": 5.141437308868502e-05,
+      "loss": 0.1197,
+      "step": 11740
+    },
+    {
+      "epoch": 107.72,
+      "learning_rate": 5.137614678899083e-05,
+      "loss": 0.0641,
+      "step": 11741
+    },
+    {
+      "epoch": 107.72,
+      "learning_rate": 5.1337920489296635e-05,
+      "loss": 0.164,
+      "step": 11742
+    },
+    {
+      "epoch": 107.73,
+      "learning_rate": 5.1299694189602447e-05,
+      "loss": 0.0825,
+      "step": 11743
+    },
+    {
+      "epoch": 107.74,
+      "learning_rate": 5.126146788990826e-05,
+      "loss": 0.0959,
+      "step": 11744
+    },
+    {
+      "epoch": 107.75,
+      "learning_rate": 5.122324159021407e-05,
+      "loss": 0.1271,
+      "step": 11745
+    },
+    {
+      "epoch": 107.76,
+      "learning_rate": 5.118501529051988e-05,
+      "loss": 0.1417,
+      "step": 11746
+    },
+    {
+      "epoch": 107.77,
+      "learning_rate": 5.114678899082569e-05,
+      "loss": 0.1477,
+      "step": 11747
+    },
+    {
+      "epoch": 107.78,
+      "learning_rate": 5.1108562691131495e-05,
+      "loss": 0.0807,
+      "step": 11748
+    },
+    {
+      "epoch": 107.79,
+      "learning_rate": 5.1070336391437306e-05,
+      "loss": 0.1875,
+      "step": 11749
+    },
+    {
+      "epoch": 107.8,
+      "learning_rate": 5.103211009174312e-05,
+      "loss": 0.1407,
+      "step": 11750
+    },
+    {
+      "epoch": 107.81,
+      "learning_rate": 5.099388379204893e-05,
+      "loss": 0.1304,
+      "step": 11751
+    },
+    {
+      "epoch": 107.82,
+      "learning_rate": 5.0955657492354746e-05,
+      "loss": 0.0641,
+      "step": 11752
+    },
+    {
+      "epoch": 107.83,
+      "learning_rate": 5.091743119266056e-05,
+      "loss": 0.08,
+      "step": 11753
+    },
+    {
+      "epoch": 107.83,
+      "learning_rate": 5.087920489296636e-05,
+      "loss": 0.1035,
+      "step": 11754
+    },
+    {
+      "epoch": 107.84,
+      "learning_rate": 5.084097859327217e-05,
+      "loss": 0.1463,
+      "step": 11755
+    },
+    {
+      "epoch": 107.85,
+      "learning_rate": 5.0802752293577984e-05,
+      "loss": 0.0854,
+      "step": 11756
+    },
+    {
+      "epoch": 107.86,
+      "learning_rate": 5.0764525993883795e-05,
+      "loss": 0.0784,
+      "step": 11757
+    },
+    {
+      "epoch": 107.87,
+      "learning_rate": 5.0726299694189606e-05,
+      "loss": 0.1003,
+      "step": 11758
+    },
+    {
+      "epoch": 107.88,
+      "learning_rate": 5.068807339449542e-05,
+      "loss": 0.0643,
+      "step": 11759
+    },
+    {
+      "epoch": 107.89,
+      "learning_rate": 5.064984709480122e-05,
+      "loss": 0.1341,
+      "step": 11760
+    },
+    {
+      "epoch": 107.9,
+      "learning_rate": 5.061162079510703e-05,
+      "loss": 0.1387,
+      "step": 11761
+    },
+    {
+      "epoch": 107.91,
+      "learning_rate": 5.057339449541284e-05,
+      "loss": 0.0687,
+      "step": 11762
+    },
+    {
+      "epoch": 107.92,
+      "learning_rate": 5.0535168195718654e-05,
+      "loss": 0.135,
+      "step": 11763
+    },
+    {
+      "epoch": 107.93,
+      "learning_rate": 5.0496941896024465e-05,
+      "loss": 0.0926,
+      "step": 11764
+    },
+    {
+      "epoch": 107.94,
+      "learning_rate": 5.0458715596330276e-05,
+      "loss": 0.066,
+      "step": 11765
+    },
+    {
+      "epoch": 107.94,
+      "learning_rate": 5.042048929663608e-05,
+      "loss": 0.1003,
+      "step": 11766
+    },
+    {
+      "epoch": 107.95,
+      "learning_rate": 5.03822629969419e-05,
+      "loss": 0.1264,
+      "step": 11767
+    },
+    {
+      "epoch": 107.96,
+      "learning_rate": 5.034403669724771e-05,
+      "loss": 0.034,
+      "step": 11768
+    },
+    {
+      "epoch": 107.97,
+      "learning_rate": 5.030581039755352e-05,
+      "loss": 0.0775,
+      "step": 11769
+    },
+    {
+      "epoch": 107.98,
+      "learning_rate": 5.026758409785933e-05,
+      "loss": 0.1243,
+      "step": 11770
+    },
+    {
+      "epoch": 107.99,
+      "learning_rate": 5.022935779816514e-05,
+      "loss": 0.0525,
+      "step": 11771
+    },
+    {
+      "epoch": 108.0,
+      "learning_rate": 5.019113149847095e-05,
+      "loss": 0.1321,
+      "step": 11772
+    },
+    {
+      "epoch": 108.01,
+      "learning_rate": 5.015290519877676e-05,
+      "loss": 0.1469,
+      "step": 11773
+    },
+    {
+      "epoch": 108.02,
+      "learning_rate": 5.011467889908257e-05,
+      "loss": 0.1507,
+      "step": 11774
+    },
+    {
+      "epoch": 108.03,
+      "learning_rate": 5.007645259938838e-05,
+      "loss": 0.1157,
+      "step": 11775
+    },
+    {
+      "epoch": 108.04,
+      "learning_rate": 5.003822629969419e-05,
+      "loss": 0.1836,
+      "step": 11776
+    },
+    {
+      "epoch": 108.05,
+      "learning_rate": 5e-05,
+      "loss": 0.1412,
+      "step": 11777
+    },
+    {
+      "epoch": 108.06,
+      "learning_rate": 4.996177370030581e-05,
+      "loss": 0.1072,
+      "step": 11778
+    },
+    {
+      "epoch": 108.06,
+      "learning_rate": 4.992354740061162e-05,
+      "loss": 0.0615,
+      "step": 11779
+    },
+    {
+      "epoch": 108.07,
+      "learning_rate": 4.988532110091743e-05,
+      "loss": 0.0912,
+      "step": 11780
+    },
+    {
+      "epoch": 108.08,
+      "learning_rate": 4.984709480122324e-05,
+      "loss": 0.056,
+      "step": 11781
+    },
+    {
+      "epoch": 108.09,
+      "learning_rate": 4.980886850152906e-05,
+      "loss": 0.1506,
+      "step": 11782
+    },
+    {
+      "epoch": 108.1,
+      "learning_rate": 4.977064220183487e-05,
+      "loss": 0.1406,
+      "step": 11783
+    },
+    {
+      "epoch": 108.11,
+      "learning_rate": 4.973241590214067e-05,
+      "loss": 0.1281,
+      "step": 11784
+    },
+    {
+      "epoch": 108.12,
+      "learning_rate": 4.9694189602446484e-05,
+      "loss": 0.0919,
+      "step": 11785
+    },
+    {
+      "epoch": 108.13,
+      "learning_rate": 4.9655963302752295e-05,
+      "loss": 0.1333,
+      "step": 11786
+    },
+    {
+      "epoch": 108.14,
+      "learning_rate": 4.9617737003058106e-05,
+      "loss": 0.0999,
+      "step": 11787
+    },
+    {
+      "epoch": 108.15,
+      "learning_rate": 4.957951070336392e-05,
+      "loss": 0.0972,
+      "step": 11788
+    },
+    {
+      "epoch": 108.16,
+      "learning_rate": 4.954128440366973e-05,
+      "loss": 0.0635,
+      "step": 11789
+    },
+    {
+      "epoch": 108.17,
+      "learning_rate": 4.950305810397553e-05,
+      "loss": 0.1252,
+      "step": 11790
+    },
+    {
+      "epoch": 108.17,
+      "learning_rate": 4.9464831804281344e-05,
+      "loss": 0.0921,
+      "step": 11791
+    },
+    {
+      "epoch": 108.18,
+      "learning_rate": 4.9426605504587155e-05,
+      "loss": 0.102,
+      "step": 11792
+    },
+    {
+      "epoch": 108.19,
+      "learning_rate": 4.9388379204892966e-05,
+      "loss": 0.0686,
+      "step": 11793
+    },
+    {
+      "epoch": 108.2,
+      "learning_rate": 4.935015290519878e-05,
+      "loss": 0.0879,
+      "step": 11794
+    },
+    {
+      "epoch": 108.21,
+      "learning_rate": 4.931192660550459e-05,
+      "loss": 0.0858,
+      "step": 11795
+    },
+    {
+      "epoch": 108.22,
+      "learning_rate": 4.927370030581039e-05,
+      "loss": 0.1057,
+      "step": 11796
+    },
+    {
+      "epoch": 108.23,
+      "learning_rate": 4.923547400611621e-05,
+      "loss": 0.1558,
+      "step": 11797
+    },
+    {
+      "epoch": 108.24,
+      "learning_rate": 4.919724770642202e-05,
+      "loss": 0.0503,
+      "step": 11798
+    },
+    {
+      "epoch": 108.25,
+      "learning_rate": 4.915902140672783e-05,
+      "loss": 0.0569,
+      "step": 11799
+    },
+    {
+      "epoch": 108.26,
+      "learning_rate": 4.912079510703364e-05,
+      "loss": 0.1542,
+      "step": 11800
+    },
+    {
+      "epoch": 108.27,
+      "learning_rate": 4.9082568807339454e-05,
+      "loss": 0.1403,
+      "step": 11801
+    },
+    {
+      "epoch": 108.28,
+      "learning_rate": 4.904434250764526e-05,
+      "loss": 0.1593,
+      "step": 11802
+    },
+    {
+      "epoch": 108.28,
+      "learning_rate": 4.900611620795107e-05,
+      "loss": 0.1755,
+      "step": 11803
+    },
+    {
+      "epoch": 108.29,
+      "learning_rate": 4.896788990825688e-05,
+      "loss": 0.1152,
+      "step": 11804
+    },
+    {
+      "epoch": 108.3,
+      "learning_rate": 4.892966360856269e-05,
+      "loss": 0.0923,
+      "step": 11805
+    },
+    {
+      "epoch": 108.31,
+      "learning_rate": 4.88914373088685e-05,
+      "loss": 0.1175,
+      "step": 11806
+    },
+    {
+      "epoch": 108.32,
+      "learning_rate": 4.8853211009174314e-05,
+      "loss": 0.085,
+      "step": 11807
+    },
+    {
+      "epoch": 108.33,
+      "learning_rate": 4.881498470948012e-05,
+      "loss": 0.0799,
+      "step": 11808
+    },
+    {
+      "epoch": 108.34,
+      "learning_rate": 4.877675840978593e-05,
+      "loss": 0.1805,
+      "step": 11809
+    },
+    {
+      "epoch": 108.35,
+      "learning_rate": 4.873853211009174e-05,
+      "loss": 0.112,
+      "step": 11810
+    },
+    {
+      "epoch": 108.36,
+      "learning_rate": 4.870030581039756e-05,
+      "loss": 0.1682,
+      "step": 11811
+    },
+    {
+      "epoch": 108.37,
+      "learning_rate": 4.866207951070337e-05,
+      "loss": 0.1274,
+      "step": 11812
+    },
+    {
+      "epoch": 108.38,
+      "learning_rate": 4.862385321100918e-05,
+      "loss": 0.1297,
+      "step": 11813
+    },
+    {
+      "epoch": 108.39,
+      "learning_rate": 4.8585626911314985e-05,
+      "loss": 0.1578,
+      "step": 11814
+    },
+    {
+      "epoch": 108.39,
+      "learning_rate": 4.8547400611620796e-05,
+      "loss": 0.1241,
+      "step": 11815
+    },
+    {
+      "epoch": 108.4,
+      "learning_rate": 4.850917431192661e-05,
+      "loss": 0.0552,
+      "step": 11816
+    },
+    {
+      "epoch": 108.41,
+      "learning_rate": 4.847094801223242e-05,
+      "loss": 0.086,
+      "step": 11817
+    },
+    {
+      "epoch": 108.42,
+      "learning_rate": 4.843272171253823e-05,
+      "loss": 0.06,
+      "step": 11818
+    },
+    {
+      "epoch": 108.43,
+      "learning_rate": 4.839449541284404e-05,
+      "loss": 0.0909,
+      "step": 11819
+    },
+    {
+      "epoch": 108.44,
+      "learning_rate": 4.8356269113149844e-05,
+      "loss": 0.0681,
+      "step": 11820
+    },
+    {
+      "epoch": 108.45,
+      "learning_rate": 4.8318042813455655e-05,
+      "loss": 0.0932,
+      "step": 11821
+    },
+    {
+      "epoch": 108.46,
+      "learning_rate": 4.8279816513761466e-05,
+      "loss": 0.0535,
+      "step": 11822
+    },
+    {
+      "epoch": 108.47,
+      "learning_rate": 4.824159021406728e-05,
+      "loss": 0.1123,
+      "step": 11823
+    },
+    {
+      "epoch": 108.48,
+      "learning_rate": 4.820336391437309e-05,
+      "loss": 0.0653,
+      "step": 11824
+    },
+    {
+      "epoch": 108.49,
+      "learning_rate": 4.81651376146789e-05,
+      "loss": 0.1182,
+      "step": 11825
+    },
+    {
+      "epoch": 108.5,
+      "learning_rate": 4.812691131498471e-05,
+      "loss": 0.0623,
+      "step": 11826
+    },
+    {
+      "epoch": 108.5,
+      "learning_rate": 4.808868501529052e-05,
+      "loss": 0.145,
+      "step": 11827
+    },
+    {
+      "epoch": 108.51,
+      "learning_rate": 4.805045871559633e-05,
+      "loss": 0.1242,
+      "step": 11828
+    },
+    {
+      "epoch": 108.52,
+      "learning_rate": 4.8012232415902144e-05,
+      "loss": 0.1009,
+      "step": 11829
+    },
+    {
+      "epoch": 108.53,
+      "learning_rate": 4.7974006116207955e-05,
+      "loss": 0.0781,
+      "step": 11830
+    },
+    {
+      "epoch": 108.54,
+      "learning_rate": 4.7935779816513766e-05,
+      "loss": 0.1108,
+      "step": 11831
+    },
+    {
+      "epoch": 108.55,
+      "learning_rate": 4.789755351681957e-05,
+      "loss": 0.1157,
+      "step": 11832
+    },
+    {
+      "epoch": 108.56,
+      "learning_rate": 4.785932721712538e-05,
+      "loss": 0.073,
+      "step": 11833
+    },
+    {
+      "epoch": 108.57,
+      "learning_rate": 4.782110091743119e-05,
+      "loss": 0.1351,
+      "step": 11834
+    },
+    {
+      "epoch": 108.58,
+      "learning_rate": 4.7782874617737003e-05,
+      "loss": 0.0899,
+      "step": 11835
+    },
+    {
+      "epoch": 108.59,
+      "learning_rate": 4.7744648318042815e-05,
+      "loss": 0.1105,
+      "step": 11836
+    },
+    {
+      "epoch": 108.6,
+      "learning_rate": 4.7706422018348626e-05,
+      "loss": 0.0915,
+      "step": 11837
+    },
+    {
+      "epoch": 108.61,
+      "learning_rate": 4.766819571865443e-05,
+      "loss": 0.0895,
+      "step": 11838
+    },
+    {
+      "epoch": 108.61,
+      "learning_rate": 4.762996941896024e-05,
+      "loss": 0.0877,
+      "step": 11839
+    },
+    {
+      "epoch": 108.62,
+      "learning_rate": 4.759174311926605e-05,
+      "loss": 0.1515,
+      "step": 11840
+    },
+    {
+      "epoch": 108.63,
+      "learning_rate": 4.755351681957187e-05,
+      "loss": 0.1047,
+      "step": 11841
+    },
+    {
+      "epoch": 108.64,
+      "learning_rate": 4.751529051987768e-05,
+      "loss": 0.0695,
+      "step": 11842
+    },
+    {
+      "epoch": 108.65,
+      "learning_rate": 4.747706422018349e-05,
+      "loss": 0.1372,
+      "step": 11843
+    },
+    {
+      "epoch": 108.66,
+      "learning_rate": 4.7438837920489296e-05,
+      "loss": 0.1035,
+      "step": 11844
+    },
+    {
+      "epoch": 108.67,
+      "learning_rate": 4.740061162079511e-05,
+      "loss": 0.0471,
+      "step": 11845
+    },
+    {
+      "epoch": 108.68,
+      "learning_rate": 4.736238532110092e-05,
+      "loss": 0.1052,
+      "step": 11846
+    },
+    {
+      "epoch": 108.69,
+      "learning_rate": 4.732415902140673e-05,
+      "loss": 0.0895,
+      "step": 11847
+    },
+    {
+      "epoch": 108.7,
+      "learning_rate": 4.728593272171254e-05,
+      "loss": 0.0953,
+      "step": 11848
+    },
+    {
+      "epoch": 108.71,
+      "learning_rate": 4.724770642201835e-05,
+      "loss": 0.0429,
+      "step": 11849
+    },
+    {
+      "epoch": 108.72,
+      "learning_rate": 4.7209480122324156e-05,
+      "loss": 0.0504,
+      "step": 11850
+    },
+    {
+      "epoch": 108.72,
+      "learning_rate": 4.717125382262997e-05,
+      "loss": 0.0735,
+      "step": 11851
+    },
+    {
+      "epoch": 108.73,
+      "learning_rate": 4.713302752293578e-05,
+      "loss": 0.0891,
+      "step": 11852
+    },
+    {
+      "epoch": 108.74,
+      "learning_rate": 4.709480122324159e-05,
+      "loss": 0.0201,
+      "step": 11853
+    },
+    {
+      "epoch": 108.75,
+      "learning_rate": 4.70565749235474e-05,
+      "loss": 0.1237,
+      "step": 11854
+    },
+    {
+      "epoch": 108.76,
+      "learning_rate": 4.701834862385322e-05,
+      "loss": 0.1933,
+      "step": 11855
+    },
+    {
+      "epoch": 108.77,
+      "learning_rate": 4.698012232415902e-05,
+      "loss": 0.1287,
+      "step": 11856
+    },
+    {
+      "epoch": 108.78,
+      "learning_rate": 4.694189602446483e-05,
+      "loss": 0.0688,
+      "step": 11857
+    },
+    {
+      "epoch": 108.79,
+      "learning_rate": 4.6903669724770644e-05,
+      "loss": 0.1254,
+      "step": 11858
+    },
+    {
+      "epoch": 108.8,
+      "learning_rate": 4.6865443425076455e-05,
+      "loss": 0.1274,
+      "step": 11859
+    },
+    {
+      "epoch": 108.81,
+      "learning_rate": 4.6827217125382266e-05,
+      "loss": 0.1463,
+      "step": 11860
+    },
+    {
+      "epoch": 108.82,
+      "learning_rate": 4.678899082568808e-05,
+      "loss": 0.0692,
+      "step": 11861
+    },
+    {
+      "epoch": 108.83,
+      "learning_rate": 4.675076452599388e-05,
+      "loss": 0.1179,
+      "step": 11862
+    },
+    {
+      "epoch": 108.83,
+      "learning_rate": 4.671253822629969e-05,
+      "loss": 0.1101,
+      "step": 11863
+    },
+    {
+      "epoch": 108.84,
+      "learning_rate": 4.6674311926605504e-05,
+      "loss": 0.0844,
+      "step": 11864
+    },
+    {
+      "epoch": 108.85,
+      "learning_rate": 4.6636085626911315e-05,
+      "loss": 0.1352,
+      "step": 11865
+    },
+    {
+      "epoch": 108.86,
+      "learning_rate": 4.6597859327217126e-05,
+      "loss": 0.091,
+      "step": 11866
+    },
+    {
+      "epoch": 108.87,
+      "learning_rate": 4.655963302752294e-05,
+      "loss": 0.1393,
+      "step": 11867
+    },
+    {
+      "epoch": 108.88,
+      "learning_rate": 4.652140672782874e-05,
+      "loss": 0.092,
+      "step": 11868
+    },
+    {
+      "epoch": 108.89,
+      "learning_rate": 4.648318042813455e-05,
+      "loss": 0.1015,
+      "step": 11869
+    },
+    {
+      "epoch": 108.9,
+      "learning_rate": 4.644495412844037e-05,
+      "loss": 0.1395,
+      "step": 11870
+    },
+    {
+      "epoch": 108.91,
+      "learning_rate": 4.640672782874618e-05,
+      "loss": 0.097,
+      "step": 11871
+    },
+    {
+      "epoch": 108.92,
+      "learning_rate": 4.636850152905199e-05,
+      "loss": 0.0681,
+      "step": 11872
+    },
+    {
+      "epoch": 108.93,
+      "learning_rate": 4.6330275229357804e-05,
+      "loss": 0.1476,
+      "step": 11873
+    },
+    {
+      "epoch": 108.94,
+      "learning_rate": 4.629204892966361e-05,
+      "loss": 0.0618,
+      "step": 11874
+    },
+    {
+      "epoch": 108.94,
+      "learning_rate": 4.625382262996942e-05,
+      "loss": 0.1395,
+      "step": 11875
+    },
+    {
+      "epoch": 108.95,
+      "learning_rate": 4.621559633027523e-05,
+      "loss": 0.0913,
+      "step": 11876
+    },
+    {
+      "epoch": 108.96,
+      "learning_rate": 4.617737003058104e-05,
+      "loss": 0.1167,
+      "step": 11877
+    },
+    {
+      "epoch": 108.97,
+      "learning_rate": 4.613914373088685e-05,
+      "loss": 0.0848,
+      "step": 11878
+    },
+    {
+      "epoch": 108.98,
+      "learning_rate": 4.610091743119266e-05,
+      "loss": 0.0731,
+      "step": 11879
+    },
+    {
+      "epoch": 108.99,
+      "learning_rate": 4.606269113149847e-05,
+      "loss": 0.0943,
+      "step": 11880
+    },
+    {
+      "epoch": 109.0,
+      "learning_rate": 4.602446483180428e-05,
+      "loss": 0.065,
+      "step": 11881
+    },
+    {
+      "epoch": 109.01,
+      "learning_rate": 4.598623853211009e-05,
+      "loss": 0.1668,
+      "step": 11882
+    },
+    {
+      "epoch": 109.02,
+      "learning_rate": 4.59480122324159e-05,
+      "loss": 0.1344,
+      "step": 11883
+    },
+    {
+      "epoch": 109.03,
+      "learning_rate": 4.590978593272171e-05,
+      "loss": 0.103,
+      "step": 11884
+    },
+    {
+      "epoch": 109.04,
+      "learning_rate": 4.587155963302753e-05,
+      "loss": 0.0984,
+      "step": 11885
+    },
+    {
+      "epoch": 109.05,
+      "learning_rate": 4.5833333333333334e-05,
+      "loss": 0.094,
+      "step": 11886
+    },
+    {
+      "epoch": 109.06,
+      "learning_rate": 4.5795107033639145e-05,
+      "loss": 0.0916,
+      "step": 11887
+    },
+    {
+      "epoch": 109.06,
+      "learning_rate": 4.5756880733944956e-05,
+      "loss": 0.1134,
+      "step": 11888
+    },
+    {
+      "epoch": 109.07,
+      "learning_rate": 4.571865443425077e-05,
+      "loss": 0.1099,
+      "step": 11889
+    },
+    {
+      "epoch": 109.08,
+      "learning_rate": 4.568042813455658e-05,
+      "loss": 0.0773,
+      "step": 11890
+    },
+    {
+      "epoch": 109.09,
+      "learning_rate": 4.564220183486239e-05,
+      "loss": 0.0631,
+      "step": 11891
+    },
+    {
+      "epoch": 109.1,
+      "learning_rate": 4.5603975535168193e-05,
+      "loss": 0.1466,
+      "step": 11892
+    },
+    {
+      "epoch": 109.11,
+      "learning_rate": 4.5565749235474005e-05,
+      "loss": 0.1454,
+      "step": 11893
+    },
+    {
+      "epoch": 109.12,
+      "learning_rate": 4.5527522935779816e-05,
+      "loss": 0.0999,
+      "step": 11894
+    },
+    {
+      "epoch": 109.13,
+      "learning_rate": 4.548929663608563e-05,
+      "loss": 0.1116,
+      "step": 11895
+    },
+    {
+      "epoch": 109.14,
+      "learning_rate": 4.545107033639144e-05,
+      "loss": 0.0946,
+      "step": 11896
+    },
+    {
+      "epoch": 109.15,
+      "learning_rate": 4.541284403669725e-05,
+      "loss": 0.0894,
+      "step": 11897
+    },
+    {
+      "epoch": 109.16,
+      "learning_rate": 4.537461773700305e-05,
+      "loss": 0.0626,
+      "step": 11898
+    },
+    {
+      "epoch": 109.17,
+      "learning_rate": 4.5336391437308864e-05,
+      "loss": 0.0754,
+      "step": 11899
+    },
+    {
+      "epoch": 109.17,
+      "learning_rate": 4.529816513761468e-05,
+      "loss": 0.0901,
+      "step": 11900
+    },
+    {
+      "epoch": 109.18,
+      "learning_rate": 4.525993883792049e-05,
+      "loss": 0.0833,
+      "step": 11901
+    },
+    {
+      "epoch": 109.19,
+      "learning_rate": 4.5221712538226304e-05,
+      "loss": 0.0764,
+      "step": 11902
+    },
+    {
+      "epoch": 109.2,
+      "learning_rate": 4.5183486238532115e-05,
+      "loss": 0.0723,
+      "step": 11903
+    },
+    {
+      "epoch": 109.21,
+      "learning_rate": 4.514525993883792e-05,
+      "loss": 0.1117,
+      "step": 11904
+    },
+    {
+      "epoch": 109.22,
+      "learning_rate": 4.510703363914373e-05,
+      "loss": 0.0594,
+      "step": 11905
+    },
+    {
+      "epoch": 109.23,
+      "learning_rate": 4.506880733944954e-05,
+      "loss": 0.0856,
+      "step": 11906
+    },
+    {
+      "epoch": 109.24,
+      "learning_rate": 4.503058103975535e-05,
+      "loss": 0.0778,
+      "step": 11907
+    },
+    {
+      "epoch": 109.25,
+      "learning_rate": 4.4992354740061164e-05,
+      "loss": 0.0451,
+      "step": 11908
+    },
+    {
+      "epoch": 109.26,
+      "learning_rate": 4.4954128440366975e-05,
+      "loss": 0.1649,
+      "step": 11909
+    },
+    {
+      "epoch": 109.27,
+      "learning_rate": 4.491590214067278e-05,
+      "loss": 0.1433,
+      "step": 11910
+    },
+    {
+      "epoch": 109.28,
+      "learning_rate": 4.487767584097859e-05,
+      "loss": 0.1255,
+      "step": 11911
+    },
+    {
+      "epoch": 109.28,
+      "learning_rate": 4.48394495412844e-05,
+      "loss": 0.1621,
+      "step": 11912
+    },
+    {
+      "epoch": 109.29,
+      "learning_rate": 4.480122324159021e-05,
+      "loss": 0.0954,
+      "step": 11913
+    },
+    {
+      "epoch": 109.3,
+      "learning_rate": 4.476299694189603e-05,
+      "loss": 0.0942,
+      "step": 11914
+    },
+    {
+      "epoch": 109.31,
+      "learning_rate": 4.472477064220184e-05,
+      "loss": 0.1078,
+      "step": 11915
+    },
+    {
+      "epoch": 109.32,
+      "learning_rate": 4.4686544342507645e-05,
+      "loss": 0.0974,
+      "step": 11916
+    },
+    {
+      "epoch": 109.33,
+      "learning_rate": 4.4648318042813456e-05,
+      "loss": 0.1129,
+      "step": 11917
+    },
+    {
+      "epoch": 109.34,
+      "learning_rate": 4.461009174311927e-05,
+      "loss": 0.0888,
+      "step": 11918
+    },
+    {
+      "epoch": 109.35,
+      "learning_rate": 4.457186544342508e-05,
+      "loss": 0.1114,
+      "step": 11919
+    },
+    {
+      "epoch": 109.36,
+      "learning_rate": 4.453363914373089e-05,
+      "loss": 0.0956,
+      "step": 11920
+    },
+    {
+      "epoch": 109.37,
+      "learning_rate": 4.44954128440367e-05,
+      "loss": 0.0699,
+      "step": 11921
+    },
+    {
+      "epoch": 109.38,
+      "learning_rate": 4.4457186544342505e-05,
+      "loss": 0.1285,
+      "step": 11922
+    },
+    {
+      "epoch": 109.39,
+      "learning_rate": 4.4418960244648316e-05,
+      "loss": 0.1211,
+      "step": 11923
+    },
+    {
+      "epoch": 109.39,
+      "learning_rate": 4.438073394495413e-05,
+      "loss": 0.1342,
+      "step": 11924
+    },
+    {
+      "epoch": 109.4,
+      "learning_rate": 4.434250764525994e-05,
+      "loss": 0.1013,
+      "step": 11925
+    },
+    {
+      "epoch": 109.41,
+      "learning_rate": 4.430428134556575e-05,
+      "loss": 0.0559,
+      "step": 11926
+    },
+    {
+      "epoch": 109.42,
+      "learning_rate": 4.426605504587156e-05,
+      "loss": 0.1485,
+      "step": 11927
+    },
+    {
+      "epoch": 109.43,
+      "learning_rate": 4.4227828746177365e-05,
+      "loss": 0.1046,
+      "step": 11928
+    },
+    {
+      "epoch": 109.44,
+      "learning_rate": 4.418960244648318e-05,
+      "loss": 0.0808,
+      "step": 11929
+    },
+    {
+      "epoch": 109.45,
+      "learning_rate": 4.4151376146788994e-05,
+      "loss": 0.0942,
+      "step": 11930
+    },
+    {
+      "epoch": 109.46,
+      "learning_rate": 4.4113149847094805e-05,
+      "loss": 0.1078,
+      "step": 11931
+    },
+    {
+      "epoch": 109.47,
+      "learning_rate": 4.4074923547400616e-05,
+      "loss": 0.0812,
+      "step": 11932
+    },
+    {
+      "epoch": 109.48,
+      "learning_rate": 4.403669724770643e-05,
+      "loss": 0.1018,
+      "step": 11933
+    },
+    {
+      "epoch": 109.49,
+      "learning_rate": 4.399847094801223e-05,
+      "loss": 0.1407,
+      "step": 11934
+    },
+    {
+      "epoch": 109.5,
+      "learning_rate": 4.396024464831804e-05,
+      "loss": 0.028,
+      "step": 11935
+    },
+    {
+      "epoch": 109.5,
+      "learning_rate": 4.392201834862385e-05,
+      "loss": 0.1415,
+      "step": 11936
+    },
+    {
+      "epoch": 109.51,
+      "learning_rate": 4.3883792048929664e-05,
+      "loss": 0.0698,
+      "step": 11937
+    },
+    {
+      "epoch": 109.52,
+      "learning_rate": 4.3845565749235475e-05,
+      "loss": 0.1368,
+      "step": 11938
+    },
+    {
+      "epoch": 109.53,
+      "learning_rate": 4.3807339449541286e-05,
+      "loss": 0.0893,
+      "step": 11939
+    },
+    {
+      "epoch": 109.54,
+      "learning_rate": 4.376911314984709e-05,
+      "loss": 0.1735,
+      "step": 11940
+    },
+    {
+      "epoch": 109.55,
+      "learning_rate": 4.37308868501529e-05,
+      "loss": 0.1256,
+      "step": 11941
+    },
+    {
+      "epoch": 109.56,
+      "learning_rate": 4.369266055045871e-05,
+      "loss": 0.1012,
+      "step": 11942
+    },
+    {
+      "epoch": 109.57,
+      "learning_rate": 4.3654434250764524e-05,
+      "loss": 0.1288,
+      "step": 11943
+    },
+    {
+      "epoch": 109.58,
+      "learning_rate": 4.361620795107034e-05,
+      "loss": 0.0989,
+      "step": 11944
+    },
+    {
+      "epoch": 109.59,
+      "learning_rate": 4.357798165137615e-05,
+      "loss": 0.1543,
+      "step": 11945
+    },
+    {
+      "epoch": 109.6,
+      "learning_rate": 4.353975535168196e-05,
+      "loss": 0.1148,
+      "step": 11946
+    },
+    {
+      "epoch": 109.61,
+      "learning_rate": 4.350152905198777e-05,
+      "loss": 0.1275,
+      "step": 11947
+    },
+    {
+      "epoch": 109.61,
+      "learning_rate": 4.346330275229358e-05,
+      "loss": 0.1021,
+      "step": 11948
+    },
+    {
+      "epoch": 109.62,
+      "learning_rate": 4.342507645259939e-05,
+      "loss": 0.1416,
+      "step": 11949
+    },
+    {
+      "epoch": 109.63,
+      "learning_rate": 4.33868501529052e-05,
+      "loss": 0.0743,
+      "step": 11950
+    },
+    {
+      "epoch": 109.64,
+      "learning_rate": 4.334862385321101e-05,
+      "loss": 0.1471,
+      "step": 11951
+    },
+    {
+      "epoch": 109.65,
+      "learning_rate": 4.331039755351682e-05,
+      "loss": 0.0726,
+      "step": 11952
+    },
+    {
+      "epoch": 109.66,
+      "learning_rate": 4.327217125382263e-05,
+      "loss": 0.1197,
+      "step": 11953
+    },
+    {
+      "epoch": 109.67,
+      "learning_rate": 4.323394495412844e-05,
+      "loss": 0.0686,
+      "step": 11954
+    },
+    {
+      "epoch": 109.68,
+      "learning_rate": 4.319571865443425e-05,
+      "loss": 0.074,
+      "step": 11955
+    },
+    {
+      "epoch": 109.69,
+      "learning_rate": 4.315749235474006e-05,
+      "loss": 0.1093,
+      "step": 11956
+    },
+    {
+      "epoch": 109.7,
+      "learning_rate": 4.311926605504587e-05,
+      "loss": 0.0747,
+      "step": 11957
+    },
+    {
+      "epoch": 109.71,
+      "learning_rate": 4.3081039755351676e-05,
+      "loss": 0.0691,
+      "step": 11958
+    },
+    {
+      "epoch": 109.72,
+      "learning_rate": 4.3042813455657494e-05,
+      "loss": 0.0473,
+      "step": 11959
+    },
+    {
+      "epoch": 109.72,
+      "learning_rate": 4.3004587155963305e-05,
+      "loss": 0.0604,
+      "step": 11960
+    },
+    {
+      "epoch": 109.73,
+      "learning_rate": 4.2966360856269116e-05,
+      "loss": 0.117,
+      "step": 11961
+    },
+    {
+      "epoch": 109.74,
+      "learning_rate": 4.292813455657493e-05,
+      "loss": 0.0368,
+      "step": 11962
+    },
+    {
+      "epoch": 109.75,
+      "learning_rate": 4.288990825688074e-05,
+      "loss": 0.1685,
+      "step": 11963
+    },
+    {
+      "epoch": 109.76,
+      "learning_rate": 4.285168195718654e-05,
+      "loss": 0.133,
+      "step": 11964
+    },
+    {
+      "epoch": 109.77,
+      "learning_rate": 4.2813455657492354e-05,
+      "loss": 0.1143,
+      "step": 11965
+    },
+    {
+      "epoch": 109.78,
+      "learning_rate": 4.2775229357798165e-05,
+      "loss": 0.1234,
+      "step": 11966
+    },
+    {
+      "epoch": 109.79,
+      "learning_rate": 4.2737003058103976e-05,
+      "loss": 0.1344,
+      "step": 11967
+    },
+    {
+      "epoch": 109.8,
+      "learning_rate": 4.269877675840979e-05,
+      "loss": 0.0817,
+      "step": 11968
+    },
+    {
+      "epoch": 109.81,
+      "learning_rate": 4.26605504587156e-05,
+      "loss": 0.1299,
+      "step": 11969
+    },
+    {
+      "epoch": 109.82,
+      "learning_rate": 4.26223241590214e-05,
+      "loss": 0.0817,
+      "step": 11970
+    },
+    {
+      "epoch": 109.83,
+      "learning_rate": 4.258409785932721e-05,
+      "loss": 0.0609,
+      "step": 11971
+    },
+    {
+      "epoch": 109.83,
+      "learning_rate": 4.2545871559633024e-05,
+      "loss": 0.0976,
+      "step": 11972
+    },
+    {
+      "epoch": 109.84,
+      "learning_rate": 4.250764525993884e-05,
+      "loss": 0.086,
+      "step": 11973
+    },
+    {
+      "epoch": 109.85,
+      "learning_rate": 4.246941896024465e-05,
+      "loss": 0.0446,
+      "step": 11974
+    },
+    {
+      "epoch": 109.86,
+      "learning_rate": 4.2431192660550464e-05,
+      "loss": 0.1066,
+      "step": 11975
+    },
+    {
+      "epoch": 109.87,
+      "learning_rate": 4.239296636085627e-05,
+      "loss": 0.1287,
+      "step": 11976
+    },
+    {
+      "epoch": 109.88,
+      "learning_rate": 4.235474006116208e-05,
+      "loss": 0.1215,
+      "step": 11977
+    },
+    {
+      "epoch": 109.89,
+      "learning_rate": 4.231651376146789e-05,
+      "loss": 0.0453,
+      "step": 11978
+    },
+    {
+      "epoch": 109.9,
+      "learning_rate": 4.22782874617737e-05,
+      "loss": 0.1052,
+      "step": 11979
+    },
+    {
+      "epoch": 109.91,
+      "learning_rate": 4.224006116207951e-05,
+      "loss": 0.0634,
+      "step": 11980
+    },
+    {
+      "epoch": 109.92,
+      "learning_rate": 4.2201834862385324e-05,
+      "loss": 0.1202,
+      "step": 11981
+    },
+    {
+      "epoch": 109.93,
+      "learning_rate": 4.216360856269113e-05,
+      "loss": 0.0915,
+      "step": 11982
+    },
+    {
+      "epoch": 109.94,
+      "learning_rate": 4.212538226299694e-05,
+      "loss": 0.0472,
+      "step": 11983
+    },
+    {
+      "epoch": 109.94,
+      "learning_rate": 4.208715596330275e-05,
+      "loss": 0.0651,
+      "step": 11984
+    },
+    {
+      "epoch": 109.95,
+      "learning_rate": 4.204892966360856e-05,
+      "loss": 0.0771,
+      "step": 11985
+    },
+    {
+      "epoch": 109.96,
+      "learning_rate": 4.201070336391437e-05,
+      "loss": 0.0904,
+      "step": 11986
+    },
+    {
+      "epoch": 109.97,
+      "learning_rate": 4.1972477064220184e-05,
+      "loss": 0.0958,
+      "step": 11987
+    },
+    {
+      "epoch": 109.98,
+      "learning_rate": 4.1934250764525995e-05,
+      "loss": 0.0862,
+      "step": 11988
+    },
+    {
+      "epoch": 109.99,
+      "learning_rate": 4.1896024464831806e-05,
+      "loss": 0.0167,
+      "step": 11989
+    },
+    {
+      "epoch": 110.0,
+      "learning_rate": 4.185779816513762e-05,
+      "loss": 0.071,
+      "step": 11990
+    },
+    {
+      "epoch": 110.01,
+      "learning_rate": 4.181957186544343e-05,
+      "loss": 0.1376,
+      "step": 11991
+    },
+    {
+      "epoch": 110.02,
+      "learning_rate": 4.178134556574924e-05,
+      "loss": 0.1091,
+      "step": 11992
+    },
+    {
+      "epoch": 110.03,
+      "learning_rate": 4.174311926605505e-05,
+      "loss": 0.0853,
+      "step": 11993
+    },
+    {
+      "epoch": 110.04,
+      "learning_rate": 4.1704892966360854e-05,
+      "loss": 0.095,
+      "step": 11994
+    },
+    {
+      "epoch": 110.05,
+      "learning_rate": 4.1666666666666665e-05,
+      "loss": 0.118,
+      "step": 11995
+    },
+    {
+      "epoch": 110.06,
+      "learning_rate": 4.1628440366972476e-05,
+      "loss": 0.0885,
+      "step": 11996
+    },
+    {
+      "epoch": 110.06,
+      "learning_rate": 4.159021406727829e-05,
+      "loss": 0.0897,
+      "step": 11997
+    },
+    {
+      "epoch": 110.07,
+      "learning_rate": 4.15519877675841e-05,
+      "loss": 0.0928,
+      "step": 11998
+    },
+    {
+      "epoch": 110.08,
+      "learning_rate": 4.151376146788991e-05,
+      "loss": 0.1192,
+      "step": 11999
+    },
+    {
+      "epoch": 110.09,
+      "learning_rate": 4.147553516819572e-05,
+      "loss": 0.0963,
+      "step": 12000
+    },
+    {
+      "epoch": 110.09,
+      "eval_cer": 0.13617895320769427,
+      "eval_loss": 0.7642955183982849,
+      "eval_runtime": 87.0494,
+      "eval_samples_per_second": 18.92,
+      "eval_steps_per_second": 2.366,
+      "eval_wer": 0.47594729853947504,
+      "step": 12000
+    },
+    {
+      "epoch": 110.1,
+      "learning_rate": 4.1437308868501525e-05,
+      "loss": 0.1475,
+      "step": 12001
+    },
+    {
+      "epoch": 110.11,
+      "learning_rate": 4.1399082568807336e-05,
+      "loss": 0.1015,
+      "step": 12002
+    },
+    {
+      "epoch": 110.12,
+      "learning_rate": 4.1360856269113154e-05,
+      "loss": 0.0755,
+      "step": 12003
+    },
+    {
+      "epoch": 110.13,
+      "learning_rate": 4.1322629969418965e-05,
+      "loss": 0.0998,
+      "step": 12004
+    },
+    {
+      "epoch": 110.14,
+      "learning_rate": 4.1284403669724776e-05,
+      "loss": 0.0921,
+      "step": 12005
+    },
+    {
+      "epoch": 110.15,
+      "learning_rate": 4.124617737003059e-05,
+      "loss": 0.1406,
+      "step": 12006
+    },
+    {
+      "epoch": 110.16,
+      "learning_rate": 4.120795107033639e-05,
+      "loss": 0.0397,
+      "step": 12007
+    },
+    {
+      "epoch": 110.17,
+      "learning_rate": 4.11697247706422e-05,
+      "loss": 0.1278,
+      "step": 12008
+    },
+    {
+      "epoch": 110.17,
+      "learning_rate": 4.1131498470948013e-05,
+      "loss": 0.058,
+      "step": 12009
+    },
+    {
+      "epoch": 110.18,
+      "learning_rate": 4.1093272171253824e-05,
+      "loss": 0.0964,
+      "step": 12010
+    },
+    {
+      "epoch": 110.19,
+      "learning_rate": 4.1055045871559636e-05,
+      "loss": 0.0539,
+      "step": 12011
+    },
+    {
+      "epoch": 110.2,
+      "learning_rate": 4.1016819571865447e-05,
+      "loss": 0.0817,
+      "step": 12012
+    },
+    {
+      "epoch": 110.21,
+      "learning_rate": 4.097859327217125e-05,
+      "loss": 0.1097,
+      "step": 12013
+    },
+    {
+      "epoch": 110.22,
+      "learning_rate": 4.094036697247706e-05,
+      "loss": 0.0749,
+      "step": 12014
+    },
+    {
+      "epoch": 110.23,
+      "learning_rate": 4.090214067278287e-05,
+      "loss": 0.0943,
+      "step": 12015
+    },
+    {
+      "epoch": 110.24,
+      "learning_rate": 4.0863914373088684e-05,
+      "loss": 0.0617,
+      "step": 12016
+    },
+    {
+      "epoch": 110.25,
+      "learning_rate": 4.08256880733945e-05,
+      "loss": 0.0748,
+      "step": 12017
+    },
+    {
+      "epoch": 110.26,
+      "learning_rate": 4.078746177370031e-05,
+      "loss": 0.1626,
+      "step": 12018
+    },
+    {
+      "epoch": 110.27,
+      "learning_rate": 4.074923547400612e-05,
+      "loss": 0.1561,
+      "step": 12019
+    },
+    {
+      "epoch": 110.28,
+      "learning_rate": 4.071100917431193e-05,
+      "loss": 0.0964,
+      "step": 12020
+    },
+    {
+      "epoch": 110.28,
+      "learning_rate": 4.067278287461774e-05,
+      "loss": 0.1187,
+      "step": 12021
+    },
+    {
+      "epoch": 110.29,
+      "learning_rate": 4.063455657492355e-05,
+      "loss": 0.1173,
+      "step": 12022
+    },
+    {
+      "epoch": 110.3,
+      "learning_rate": 4.059633027522936e-05,
+      "loss": 0.1462,
+      "step": 12023
+    },
+    {
+      "epoch": 110.31,
+      "learning_rate": 4.055810397553517e-05,
+      "loss": 0.1629,
+      "step": 12024
+    },
+    {
+      "epoch": 110.32,
+      "learning_rate": 4.051987767584098e-05,
+      "loss": 0.0879,
+      "step": 12025
+    },
+    {
+      "epoch": 110.33,
+      "learning_rate": 4.048165137614679e-05,
+      "loss": 0.0833,
+      "step": 12026
+    },
+    {
+      "epoch": 110.34,
+      "learning_rate": 4.04434250764526e-05,
+      "loss": 0.0976,
+      "step": 12027
+    },
+    {
+      "epoch": 110.35,
+      "learning_rate": 4.040519877675841e-05,
+      "loss": 0.1179,
+      "step": 12028
+    },
+    {
+      "epoch": 110.36,
+      "learning_rate": 4.036697247706422e-05,
+      "loss": 0.1248,
+      "step": 12029
+    },
+    {
+      "epoch": 110.37,
+      "learning_rate": 4.032874617737003e-05,
+      "loss": 0.0732,
+      "step": 12030
+    },
+    {
+      "epoch": 110.38,
+      "learning_rate": 4.0290519877675836e-05,
+      "loss": 0.099,
+      "step": 12031
+    },
+    {
+      "epoch": 110.39,
+      "learning_rate": 4.0252293577981654e-05,
+      "loss": 0.0831,
+      "step": 12032
+    },
+    {
+      "epoch": 110.39,
+      "learning_rate": 4.0214067278287465e-05,
+      "loss": 0.0919,
+      "step": 12033
+    },
+    {
+      "epoch": 110.4,
+      "learning_rate": 4.0175840978593276e-05,
+      "loss": 0.0738,
+      "step": 12034
+    },
+    {
+      "epoch": 110.41,
+      "learning_rate": 4.013761467889909e-05,
+      "loss": 0.0434,
+      "step": 12035
+    },
+    {
+      "epoch": 110.42,
+      "learning_rate": 4.00993883792049e-05,
+      "loss": 0.0969,
+      "step": 12036
+    },
+    {
+      "epoch": 110.43,
+      "learning_rate": 4.00611620795107e-05,
+      "loss": 0.0819,
+      "step": 12037
+    },
+    {
+      "epoch": 110.44,
+      "learning_rate": 4.0022935779816514e-05,
+      "loss": 0.116,
+      "step": 12038
+    },
+    {
+      "epoch": 110.45,
+      "learning_rate": 3.9984709480122325e-05,
+      "loss": 0.0745,
+      "step": 12039
+    },
+    {
+      "epoch": 110.46,
+      "learning_rate": 3.9946483180428136e-05,
+      "loss": 0.1263,
+      "step": 12040
+    },
+    {
+      "epoch": 110.47,
+      "learning_rate": 3.990825688073395e-05,
+      "loss": 0.0864,
+      "step": 12041
+    },
+    {
+      "epoch": 110.48,
+      "learning_rate": 3.987003058103976e-05,
+      "loss": 0.0272,
+      "step": 12042
+    },
+    {
+      "epoch": 110.49,
+      "learning_rate": 3.983180428134556e-05,
+      "loss": 0.1005,
+      "step": 12043
+    },
+    {
+      "epoch": 110.5,
+      "learning_rate": 3.9793577981651374e-05,
+      "loss": 0.0653,
+      "step": 12044
+    },
+    {
+      "epoch": 110.5,
+      "learning_rate": 3.9755351681957185e-05,
+      "loss": 0.142,
+      "step": 12045
+    },
+    {
+      "epoch": 110.51,
+      "learning_rate": 3.9717125382262996e-05,
+      "loss": 0.1016,
+      "step": 12046
+    },
+    {
+      "epoch": 110.52,
+      "learning_rate": 3.9678899082568813e-05,
+      "loss": 0.1055,
+      "step": 12047
+    },
+    {
+      "epoch": 110.53,
+      "learning_rate": 3.9640672782874625e-05,
+      "loss": 0.0697,
+      "step": 12048
+    },
+    {
+      "epoch": 110.54,
+      "learning_rate": 3.960244648318043e-05,
+      "loss": 0.107,
+      "step": 12049
+    },
+    {
+      "epoch": 110.55,
+      "learning_rate": 3.956422018348624e-05,
+      "loss": 0.1254,
+      "step": 12050
+    },
+    {
+      "epoch": 110.56,
+      "learning_rate": 3.952599388379205e-05,
+      "loss": 0.1093,
+      "step": 12051
+    },
+    {
+      "epoch": 110.57,
+      "learning_rate": 3.948776758409786e-05,
+      "loss": 0.1121,
+      "step": 12052
+    },
+    {
+      "epoch": 110.58,
+      "learning_rate": 3.944954128440367e-05,
+      "loss": 0.1047,
+      "step": 12053
+    },
+    {
+      "epoch": 110.59,
+      "learning_rate": 3.9411314984709484e-05,
+      "loss": 0.1453,
+      "step": 12054
+    },
+    {
+      "epoch": 110.6,
+      "learning_rate": 3.937308868501529e-05,
+      "loss": 0.1033,
+      "step": 12055
+    },
+    {
+      "epoch": 110.61,
+      "learning_rate": 3.93348623853211e-05,
+      "loss": 0.0806,
+      "step": 12056
+    },
+    {
+      "epoch": 110.61,
+      "learning_rate": 3.929663608562691e-05,
+      "loss": 0.1124,
+      "step": 12057
+    },
+    {
+      "epoch": 110.62,
+      "learning_rate": 3.925840978593272e-05,
+      "loss": 0.1152,
+      "step": 12058
+    },
+    {
+      "epoch": 110.63,
+      "learning_rate": 3.922018348623853e-05,
+      "loss": 0.1285,
+      "step": 12059
+    },
+    {
+      "epoch": 110.64,
+      "learning_rate": 3.9181957186544344e-05,
+      "loss": 0.0902,
+      "step": 12060
+    },
+    {
+      "epoch": 110.65,
+      "learning_rate": 3.914373088685015e-05,
+      "loss": 0.0497,
+      "step": 12061
+    },
+    {
+      "epoch": 110.66,
+      "learning_rate": 3.9105504587155966e-05,
+      "loss": 0.0456,
+      "step": 12062
+    },
+    {
+      "epoch": 110.67,
+      "learning_rate": 3.906727828746178e-05,
+      "loss": 0.067,
+      "step": 12063
+    },
+    {
+      "epoch": 110.68,
+      "learning_rate": 3.902905198776759e-05,
+      "loss": 0.0746,
+      "step": 12064
+    },
+    {
+      "epoch": 110.69,
+      "learning_rate": 3.89908256880734e-05,
+      "loss": 0.1191,
+      "step": 12065
+    },
+    {
+      "epoch": 110.7,
+      "learning_rate": 3.895259938837921e-05,
+      "loss": 0.0754,
+      "step": 12066
+    },
+    {
+      "epoch": 110.71,
+      "learning_rate": 3.8914373088685014e-05,
+      "loss": 0.0825,
+      "step": 12067
+    },
+    {
+      "epoch": 110.72,
+      "learning_rate": 3.8876146788990826e-05,
+      "loss": 0.1,
+      "step": 12068
+    },
+    {
+      "epoch": 110.72,
+      "learning_rate": 3.8837920489296637e-05,
+      "loss": 0.1143,
+      "step": 12069
+    },
+    {
+      "epoch": 110.73,
+      "learning_rate": 3.879969418960245e-05,
+      "loss": 0.082,
+      "step": 12070
+    },
+    {
+      "epoch": 110.74,
+      "learning_rate": 3.876146788990826e-05,
+      "loss": 0.0589,
+      "step": 12071
+    },
+    {
+      "epoch": 110.75,
+      "learning_rate": 3.872324159021407e-05,
+      "loss": 0.1245,
+      "step": 12072
+    },
+    {
+      "epoch": 110.76,
+      "learning_rate": 3.8685015290519874e-05,
+      "loss": 0.1354,
+      "step": 12073
+    },
+    {
+      "epoch": 110.77,
+      "learning_rate": 3.8646788990825685e-05,
+      "loss": 0.1358,
+      "step": 12074
+    },
+    {
+      "epoch": 110.78,
+      "learning_rate": 3.8608562691131496e-05,
+      "loss": 0.152,
+      "step": 12075
+    },
+    {
+      "epoch": 110.79,
+      "learning_rate": 3.8570336391437314e-05,
+      "loss": 0.1039,
+      "step": 12076
+    },
+    {
+      "epoch": 110.8,
+      "learning_rate": 3.8532110091743125e-05,
+      "loss": 0.1001,
+      "step": 12077
+    },
+    {
+      "epoch": 110.81,
+      "learning_rate": 3.8493883792048936e-05,
+      "loss": 0.1283,
+      "step": 12078
+    },
+    {
+      "epoch": 110.82,
+      "learning_rate": 3.845565749235474e-05,
+      "loss": 0.0713,
+      "step": 12079
+    },
+    {
+      "epoch": 110.83,
+      "learning_rate": 3.841743119266055e-05,
+      "loss": 0.0436,
+      "step": 12080
+    },
+    {
+      "epoch": 110.83,
+      "learning_rate": 3.837920489296636e-05,
+      "loss": 0.0598,
+      "step": 12081
+    },
+    {
+      "epoch": 110.84,
+      "learning_rate": 3.8340978593272174e-05,
+      "loss": 0.0776,
+      "step": 12082
+    },
+    {
+      "epoch": 110.85,
+      "learning_rate": 3.8302752293577985e-05,
+      "loss": 0.0788,
+      "step": 12083
+    },
+    {
+      "epoch": 110.86,
+      "learning_rate": 3.8264525993883796e-05,
+      "loss": 0.1121,
+      "step": 12084
+    },
+    {
+      "epoch": 110.87,
+      "learning_rate": 3.82262996941896e-05,
+      "loss": 0.0999,
+      "step": 12085
+    },
+    {
+      "epoch": 110.88,
+      "learning_rate": 3.818807339449541e-05,
+      "loss": 0.055,
+      "step": 12086
+    },
+    {
+      "epoch": 110.89,
+      "learning_rate": 3.814984709480122e-05,
+      "loss": 0.1133,
+      "step": 12087
+    },
+    {
+      "epoch": 110.9,
+      "learning_rate": 3.811162079510703e-05,
+      "loss": 0.1086,
+      "step": 12088
+    },
+    {
+      "epoch": 110.91,
+      "learning_rate": 3.8073394495412844e-05,
+      "loss": 0.0739,
+      "step": 12089
+    },
+    {
+      "epoch": 110.92,
+      "learning_rate": 3.8035168195718655e-05,
+      "loss": 0.1251,
+      "step": 12090
+    },
+    {
+      "epoch": 110.93,
+      "learning_rate": 3.7996941896024466e-05,
+      "loss": 0.0458,
+      "step": 12091
+    },
+    {
+      "epoch": 110.94,
+      "learning_rate": 3.795871559633028e-05,
+      "loss": 0.1108,
+      "step": 12092
+    },
+    {
+      "epoch": 110.94,
+      "learning_rate": 3.792048929663609e-05,
+      "loss": 0.0746,
+      "step": 12093
+    },
+    {
+      "epoch": 110.95,
+      "learning_rate": 3.78822629969419e-05,
+      "loss": 0.0573,
+      "step": 12094
+    },
+    {
+      "epoch": 110.96,
+      "learning_rate": 3.784403669724771e-05,
+      "loss": 0.0794,
+      "step": 12095
+    },
+    {
+      "epoch": 110.97,
+      "learning_rate": 3.780581039755352e-05,
+      "loss": 0.0221,
+      "step": 12096
+    },
+    {
+      "epoch": 110.98,
+      "learning_rate": 3.7767584097859326e-05,
+      "loss": 0.0386,
+      "step": 12097
+    },
+    {
+      "epoch": 110.99,
+      "learning_rate": 3.772935779816514e-05,
+      "loss": 0.0231,
+      "step": 12098
+    },
+    {
+      "epoch": 111.0,
+      "learning_rate": 3.769113149847095e-05,
+      "loss": 0.0608,
+      "step": 12099
+    },
+    {
+      "epoch": 111.01,
+      "learning_rate": 3.765290519877676e-05,
+      "loss": 0.1555,
+      "step": 12100
+    },
+    {
+      "epoch": 111.02,
+      "learning_rate": 3.761467889908257e-05,
+      "loss": 0.1343,
+      "step": 12101
+    },
+    {
+      "epoch": 111.03,
+      "learning_rate": 3.757645259938838e-05,
+      "loss": 0.0593,
+      "step": 12102
+    },
+    {
+      "epoch": 111.04,
+      "learning_rate": 3.7538226299694186e-05,
+      "loss": 0.1083,
+      "step": 12103
+    },
+    {
+      "epoch": 111.05,
+      "learning_rate": 3.75e-05,
+      "loss": 0.117,
+      "step": 12104
+    },
+    {
+      "epoch": 111.06,
+      "learning_rate": 3.746177370030581e-05,
+      "loss": 0.1,
+      "step": 12105
+    },
+    {
+      "epoch": 111.06,
+      "learning_rate": 3.7423547400611626e-05,
+      "loss": 0.1475,
+      "step": 12106
+    },
+    {
+      "epoch": 111.07,
+      "learning_rate": 3.738532110091744e-05,
+      "loss": 0.0927,
+      "step": 12107
+    },
+    {
+      "epoch": 111.08,
+      "learning_rate": 3.734709480122325e-05,
+      "loss": 0.0752,
+      "step": 12108
+    },
+    {
+      "epoch": 111.09,
+      "learning_rate": 3.730886850152905e-05,
+      "loss": 0.1107,
+      "step": 12109
+    },
+    {
+      "epoch": 111.1,
+      "learning_rate": 3.727064220183486e-05,
+      "loss": 0.0886,
+      "step": 12110
+    },
+    {
+      "epoch": 111.11,
+      "learning_rate": 3.7232415902140674e-05,
+      "loss": 0.0738,
+      "step": 12111
+    },
+    {
+      "epoch": 111.12,
+      "learning_rate": 3.7194189602446485e-05,
+      "loss": 0.0865,
+      "step": 12112
+    },
+    {
+      "epoch": 111.13,
+      "learning_rate": 3.7155963302752296e-05,
+      "loss": 0.1331,
+      "step": 12113
+    },
+    {
+      "epoch": 111.14,
+      "learning_rate": 3.711773700305811e-05,
+      "loss": 0.1049,
+      "step": 12114
+    },
+    {
+      "epoch": 111.15,
+      "learning_rate": 3.707951070336391e-05,
+      "loss": 0.107,
+      "step": 12115
+    },
+    {
+      "epoch": 111.16,
+      "learning_rate": 3.704128440366972e-05,
+      "loss": 0.0817,
+      "step": 12116
+    },
+    {
+      "epoch": 111.17,
+      "learning_rate": 3.7003058103975534e-05,
+      "loss": 0.0754,
+      "step": 12117
+    },
+    {
+      "epoch": 111.17,
+      "learning_rate": 3.6964831804281345e-05,
+      "loss": 0.088,
+      "step": 12118
+    },
+    {
+      "epoch": 111.18,
+      "learning_rate": 3.6926605504587156e-05,
+      "loss": 0.1143,
+      "step": 12119
+    },
+    {
+      "epoch": 111.19,
+      "learning_rate": 3.688837920489297e-05,
+      "loss": 0.0458,
+      "step": 12120
+    },
+    {
+      "epoch": 111.2,
+      "learning_rate": 3.685015290519878e-05,
+      "loss": 0.1107,
+      "step": 12121
+    },
+    {
+      "epoch": 111.21,
+      "learning_rate": 3.681192660550459e-05,
+      "loss": 0.0843,
+      "step": 12122
+    },
+    {
+      "epoch": 111.22,
+      "learning_rate": 3.67737003058104e-05,
+      "loss": 0.1382,
+      "step": 12123
+    },
+    {
+      "epoch": 111.23,
+      "learning_rate": 3.673547400611621e-05,
+      "loss": 0.0658,
+      "step": 12124
+    },
+    {
+      "epoch": 111.24,
+      "learning_rate": 3.669724770642202e-05,
+      "loss": 0.0487,
+      "step": 12125
+    },
+    {
+      "epoch": 111.25,
+      "learning_rate": 3.665902140672783e-05,
+      "loss": 0.0644,
+      "step": 12126
+    },
+    {
+      "epoch": 111.26,
+      "learning_rate": 3.662079510703364e-05,
+      "loss": 0.1312,
+      "step": 12127
+    },
+    {
+      "epoch": 111.27,
+      "learning_rate": 3.658256880733945e-05,
+      "loss": 0.1026,
+      "step": 12128
+    },
+    {
+      "epoch": 111.28,
+      "learning_rate": 3.654434250764526e-05,
+      "loss": 0.0774,
+      "step": 12129
+    },
+    {
+      "epoch": 111.28,
+      "learning_rate": 3.650611620795107e-05,
+      "loss": 0.1157,
+      "step": 12130
+    },
+    {
+      "epoch": 111.29,
+      "learning_rate": 3.646788990825688e-05,
+      "loss": 0.0978,
+      "step": 12131
+    },
+    {
+      "epoch": 111.3,
+      "learning_rate": 3.642966360856269e-05,
+      "loss": 0.0947,
+      "step": 12132
+    },
+    {
+      "epoch": 111.31,
+      "learning_rate": 3.63914373088685e-05,
+      "loss": 0.0743,
+      "step": 12133
+    },
+    {
+      "epoch": 111.32,
+      "learning_rate": 3.635321100917431e-05,
+      "loss": 0.0908,
+      "step": 12134
+    },
+    {
+      "epoch": 111.33,
+      "learning_rate": 3.6314984709480126e-05,
+      "loss": 0.0603,
+      "step": 12135
+    },
+    {
+      "epoch": 111.34,
+      "learning_rate": 3.627675840978594e-05,
+      "loss": 0.0887,
+      "step": 12136
+    },
+    {
+      "epoch": 111.35,
+      "learning_rate": 3.623853211009175e-05,
+      "loss": 0.065,
+      "step": 12137
+    },
+    {
+      "epoch": 111.36,
+      "learning_rate": 3.620030581039756e-05,
+      "loss": 0.1628,
+      "step": 12138
+    },
+    {
+      "epoch": 111.37,
+      "learning_rate": 3.6162079510703364e-05,
+      "loss": 0.0864,
+      "step": 12139
+    },
+    {
+      "epoch": 111.38,
+      "learning_rate": 3.6123853211009175e-05,
+      "loss": 0.0522,
+      "step": 12140
+    },
+    {
+      "epoch": 111.39,
+      "learning_rate": 3.6085626911314986e-05,
+      "loss": 0.0431,
+      "step": 12141
+    },
+    {
+      "epoch": 111.39,
+      "learning_rate": 3.60474006116208e-05,
+      "loss": 0.0998,
+      "step": 12142
+    },
+    {
+      "epoch": 111.4,
+      "learning_rate": 3.600917431192661e-05,
+      "loss": 0.0878,
+      "step": 12143
+    },
+    {
+      "epoch": 111.41,
+      "learning_rate": 3.597094801223242e-05,
+      "loss": 0.06,
+      "step": 12144
+    },
+    {
+      "epoch": 111.42,
+      "learning_rate": 3.593272171253822e-05,
+      "loss": 0.1133,
+      "step": 12145
+    },
+    {
+      "epoch": 111.43,
+      "learning_rate": 3.5894495412844034e-05,
+      "loss": 0.0823,
+      "step": 12146
+    },
+    {
+      "epoch": 111.44,
+      "learning_rate": 3.5856269113149845e-05,
+      "loss": 0.0693,
+      "step": 12147
+    },
+    {
+      "epoch": 111.45,
+      "learning_rate": 3.5818042813455656e-05,
+      "loss": 0.064,
+      "step": 12148
+    },
+    {
+      "epoch": 111.46,
+      "learning_rate": 3.577981651376147e-05,
+      "loss": 0.0762,
+      "step": 12149
+    },
+    {
+      "epoch": 111.47,
+      "learning_rate": 3.5741590214067285e-05,
+      "loss": 0.0828,
+      "step": 12150
+    },
+    {
+      "epoch": 111.48,
+      "learning_rate": 3.570336391437309e-05,
+      "loss": 0.1443,
+      "step": 12151
+    },
+    {
+      "epoch": 111.49,
+      "learning_rate": 3.56651376146789e-05,
+      "loss": 0.0388,
+      "step": 12152
+    },
+    {
+      "epoch": 111.5,
+      "learning_rate": 3.562691131498471e-05,
+      "loss": 0.0785,
+      "step": 12153
+    },
+    {
+      "epoch": 111.5,
+      "learning_rate": 3.558868501529052e-05,
+      "loss": 0.1751,
+      "step": 12154
+    },
+    {
+      "epoch": 111.51,
+      "learning_rate": 3.5550458715596334e-05,
+      "loss": 0.1111,
+      "step": 12155
+    },
+    {
+      "epoch": 111.52,
+      "learning_rate": 3.5512232415902145e-05,
+      "loss": 0.0921,
+      "step": 12156
+    },
+    {
+      "epoch": 111.53,
+      "learning_rate": 3.547400611620795e-05,
+      "loss": 0.1223,
+      "step": 12157
+    },
+    {
+      "epoch": 111.54,
+      "learning_rate": 3.543577981651376e-05,
+      "loss": 0.1623,
+      "step": 12158
+    },
+    {
+      "epoch": 111.55,
+      "learning_rate": 3.539755351681957e-05,
+      "loss": 0.0684,
+      "step": 12159
+    },
+    {
+      "epoch": 111.56,
+      "learning_rate": 3.535932721712538e-05,
+      "loss": 0.0786,
+      "step": 12160
+    },
+    {
+      "epoch": 111.57,
+      "learning_rate": 3.5321100917431193e-05,
+      "loss": 0.0863,
+      "step": 12161
+    },
+    {
+      "epoch": 111.58,
+      "learning_rate": 3.5282874617737005e-05,
+      "loss": 0.0535,
+      "step": 12162
+    },
+    {
+      "epoch": 111.59,
+      "learning_rate": 3.524464831804281e-05,
+      "loss": 0.0771,
+      "step": 12163
+    },
+    {
+      "epoch": 111.6,
+      "learning_rate": 3.520642201834862e-05,
+      "loss": 0.0629,
+      "step": 12164
+    },
+    {
+      "epoch": 111.61,
+      "learning_rate": 3.516819571865444e-05,
+      "loss": 0.0816,
+      "step": 12165
+    },
+    {
+      "epoch": 111.61,
+      "learning_rate": 3.512996941896025e-05,
+      "loss": 0.0892,
+      "step": 12166
+    },
+    {
+      "epoch": 111.62,
+      "learning_rate": 3.509174311926606e-05,
+      "loss": 0.1235,
+      "step": 12167
+    },
+    {
+      "epoch": 111.63,
+      "learning_rate": 3.505351681957187e-05,
+      "loss": 0.0863,
+      "step": 12168
+    },
+    {
+      "epoch": 111.64,
+      "learning_rate": 3.5015290519877675e-05,
+      "loss": 0.1336,
+      "step": 12169
+    },
+    {
+      "epoch": 111.65,
+      "learning_rate": 3.4977064220183486e-05,
+      "loss": 0.0907,
+      "step": 12170
+    },
+    {
+      "epoch": 111.66,
+      "learning_rate": 3.49388379204893e-05,
+      "loss": 0.0506,
+      "step": 12171
+    },
+    {
+      "epoch": 111.67,
+      "learning_rate": 3.490061162079511e-05,
+      "loss": 0.0655,
+      "step": 12172
+    },
+    {
+      "epoch": 111.68,
+      "learning_rate": 3.486238532110092e-05,
+      "loss": 0.0769,
+      "step": 12173
+    },
+    {
+      "epoch": 111.69,
+      "learning_rate": 3.482415902140673e-05,
+      "loss": 0.066,
+      "step": 12174
+    },
+    {
+      "epoch": 111.7,
+      "learning_rate": 3.4785932721712535e-05,
+      "loss": 0.0903,
+      "step": 12175
+    },
+    {
+      "epoch": 111.71,
+      "learning_rate": 3.4747706422018346e-05,
+      "loss": 0.0643,
+      "step": 12176
+    },
+    {
+      "epoch": 111.72,
+      "learning_rate": 3.470948012232416e-05,
+      "loss": 0.0213,
+      "step": 12177
+    },
+    {
+      "epoch": 111.72,
+      "learning_rate": 3.467125382262997e-05,
+      "loss": 0.1499,
+      "step": 12178
+    },
+    {
+      "epoch": 111.73,
+      "learning_rate": 3.463302752293578e-05,
+      "loss": 0.0437,
+      "step": 12179
+    },
+    {
+      "epoch": 111.74,
+      "learning_rate": 3.45948012232416e-05,
+      "loss": 0.042,
+      "step": 12180
+    },
+    {
+      "epoch": 111.75,
+      "learning_rate": 3.45565749235474e-05,
+      "loss": 0.1468,
+      "step": 12181
+    },
+    {
+      "epoch": 111.76,
+      "learning_rate": 3.451834862385321e-05,
+      "loss": 0.1008,
+      "step": 12182
+    },
+    {
+      "epoch": 111.77,
+      "learning_rate": 3.448012232415902e-05,
+      "loss": 0.1077,
+      "step": 12183
+    },
+    {
+      "epoch": 111.78,
+      "learning_rate": 3.4441896024464834e-05,
+      "loss": 0.0872,
+      "step": 12184
+    },
+    {
+      "epoch": 111.79,
+      "learning_rate": 3.4403669724770645e-05,
+      "loss": 0.1167,
+      "step": 12185
+    },
+    {
+      "epoch": 111.8,
+      "learning_rate": 3.4365443425076457e-05,
+      "loss": 0.0881,
+      "step": 12186
+    },
+    {
+      "epoch": 111.81,
+      "learning_rate": 3.432721712538226e-05,
+      "loss": 0.1279,
+      "step": 12187
+    },
+    {
+      "epoch": 111.82,
+      "learning_rate": 3.428899082568807e-05,
+      "loss": 0.0598,
+      "step": 12188
+    },
+    {
+      "epoch": 111.83,
+      "learning_rate": 3.425076452599388e-05,
+      "loss": 0.0881,
+      "step": 12189
+    },
+    {
+      "epoch": 111.83,
+      "learning_rate": 3.4212538226299694e-05,
+      "loss": 0.0727,
+      "step": 12190
+    },
+    {
+      "epoch": 111.84,
+      "learning_rate": 3.4174311926605505e-05,
+      "loss": 0.1043,
+      "step": 12191
+    },
+    {
+      "epoch": 111.85,
+      "learning_rate": 3.4136085626911316e-05,
+      "loss": 0.1179,
+      "step": 12192
+    },
+    {
+      "epoch": 111.86,
+      "learning_rate": 3.409785932721712e-05,
+      "loss": 0.0735,
+      "step": 12193
+    },
+    {
+      "epoch": 111.87,
+      "learning_rate": 3.405963302752294e-05,
+      "loss": 0.073,
+      "step": 12194
+    },
+    {
+      "epoch": 111.88,
+      "learning_rate": 3.402140672782875e-05,
+      "loss": 0.0572,
+      "step": 12195
+    },
+    {
+      "epoch": 111.89,
+      "learning_rate": 3.398318042813456e-05,
+      "loss": 0.1428,
+      "step": 12196
+    },
+    {
+      "epoch": 111.9,
+      "learning_rate": 3.394495412844037e-05,
+      "loss": 0.097,
+      "step": 12197
+    },
+    {
+      "epoch": 111.91,
+      "learning_rate": 3.390672782874618e-05,
+      "loss": 0.1078,
+      "step": 12198
+    },
+    {
+      "epoch": 111.92,
+      "learning_rate": 3.386850152905199e-05,
+      "loss": 0.0786,
+      "step": 12199
+    },
+    {
+      "epoch": 111.93,
+      "learning_rate": 3.38302752293578e-05,
+      "loss": 0.0793,
+      "step": 12200
+    },
+    {
+      "epoch": 111.94,
+      "learning_rate": 3.379204892966361e-05,
+      "loss": 0.0857,
+      "step": 12201
+    },
+    {
+      "epoch": 111.94,
+      "learning_rate": 3.375382262996942e-05,
+      "loss": 0.1348,
+      "step": 12202
+    },
+    {
+      "epoch": 111.95,
+      "learning_rate": 3.371559633027523e-05,
+      "loss": 0.0849,
+      "step": 12203
+    },
+    {
+      "epoch": 111.96,
+      "learning_rate": 3.367737003058104e-05,
+      "loss": 0.0795,
+      "step": 12204
+    },
+    {
+      "epoch": 111.97,
+      "learning_rate": 3.3639143730886846e-05,
+      "loss": 0.0871,
+      "step": 12205
+    },
+    {
+      "epoch": 111.98,
+      "learning_rate": 3.360091743119266e-05,
+      "loss": 0.0565,
+      "step": 12206
+    },
+    {
+      "epoch": 111.99,
+      "learning_rate": 3.356269113149847e-05,
+      "loss": 0.0347,
+      "step": 12207
+    },
+    {
+      "epoch": 112.0,
+      "learning_rate": 3.352446483180428e-05,
+      "loss": 0.0844,
+      "step": 12208
+    },
+    {
+      "epoch": 112.01,
+      "learning_rate": 3.34862385321101e-05,
+      "loss": 0.1115,
+      "step": 12209
+    },
+    {
+      "epoch": 112.02,
+      "learning_rate": 3.344801223241591e-05,
+      "loss": 0.1375,
+      "step": 12210
+    },
+    {
+      "epoch": 112.03,
+      "learning_rate": 3.340978593272171e-05,
+      "loss": 0.1042,
+      "step": 12211
+    },
+    {
+      "epoch": 112.04,
+      "learning_rate": 3.3371559633027524e-05,
+      "loss": 0.0851,
+      "step": 12212
+    },
+    {
+      "epoch": 112.05,
+      "learning_rate": 3.3333333333333335e-05,
+      "loss": 0.11,
+      "step": 12213
+    },
+    {
+      "epoch": 112.06,
+      "learning_rate": 3.3295107033639146e-05,
+      "loss": 0.1592,
+      "step": 12214
+    },
+    {
+      "epoch": 112.06,
+      "learning_rate": 3.325688073394496e-05,
+      "loss": 0.1028,
+      "step": 12215
+    },
+    {
+      "epoch": 112.07,
+      "learning_rate": 3.321865443425077e-05,
+      "loss": 0.0711,
+      "step": 12216
+    },
+    {
+      "epoch": 112.08,
+      "learning_rate": 3.318042813455657e-05,
+      "loss": 0.0831,
+      "step": 12217
+    },
+    {
+      "epoch": 112.09,
+      "learning_rate": 3.3142201834862383e-05,
+      "loss": 0.1188,
+      "step": 12218
+    },
+    {
+      "epoch": 112.1,
+      "learning_rate": 3.3103975535168195e-05,
+      "loss": 0.1118,
+      "step": 12219
+    },
+    {
+      "epoch": 112.11,
+      "learning_rate": 3.3065749235474006e-05,
+      "loss": 0.0692,
+      "step": 12220
+    },
+    {
+      "epoch": 112.12,
+      "learning_rate": 3.302752293577982e-05,
+      "loss": 0.0755,
+      "step": 12221
+    },
+    {
+      "epoch": 112.13,
+      "learning_rate": 3.298929663608563e-05,
+      "loss": 0.0919,
+      "step": 12222
+    },
+    {
+      "epoch": 112.14,
+      "learning_rate": 3.295107033639143e-05,
+      "loss": 0.0452,
+      "step": 12223
+    },
+    {
+      "epoch": 112.15,
+      "learning_rate": 3.291284403669725e-05,
+      "loss": 0.0979,
+      "step": 12224
+    },
+    {
+      "epoch": 112.16,
+      "learning_rate": 3.287461773700306e-05,
+      "loss": 0.0866,
+      "step": 12225
+    },
+    {
+      "epoch": 112.17,
+      "learning_rate": 3.283639143730887e-05,
+      "loss": 0.1058,
+      "step": 12226
+    },
+    {
+      "epoch": 112.17,
+      "learning_rate": 3.279816513761468e-05,
+      "loss": 0.0839,
+      "step": 12227
+    },
+    {
+      "epoch": 112.18,
+      "learning_rate": 3.2759938837920494e-05,
+      "loss": 0.0861,
+      "step": 12228
+    },
+    {
+      "epoch": 112.19,
+      "learning_rate": 3.27217125382263e-05,
+      "loss": 0.1004,
+      "step": 12229
+    },
+    {
+      "epoch": 112.2,
+      "learning_rate": 3.268348623853211e-05,
+      "loss": 0.0918,
+      "step": 12230
+    },
+    {
+      "epoch": 112.21,
+      "learning_rate": 3.264525993883792e-05,
+      "loss": 0.0465,
+      "step": 12231
+    },
+    {
+      "epoch": 112.22,
+      "learning_rate": 3.260703363914373e-05,
+      "loss": 0.0755,
+      "step": 12232
+    },
+    {
+      "epoch": 112.23,
+      "learning_rate": 3.256880733944954e-05,
+      "loss": 0.0354,
+      "step": 12233
+    },
+    {
+      "epoch": 112.24,
+      "learning_rate": 3.2530581039755354e-05,
+      "loss": 0.0567,
+      "step": 12234
+    },
+    {
+      "epoch": 112.25,
+      "learning_rate": 3.249235474006116e-05,
+      "loss": 0.0868,
+      "step": 12235
+    },
+    {
+      "epoch": 112.26,
+      "learning_rate": 3.245412844036697e-05,
+      "loss": 0.1814,
+      "step": 12236
+    },
+    {
+      "epoch": 112.27,
+      "learning_rate": 3.241590214067278e-05,
+      "loss": 0.1334,
+      "step": 12237
+    },
+    {
+      "epoch": 112.28,
+      "learning_rate": 3.237767584097859e-05,
+      "loss": 0.1533,
+      "step": 12238
+    },
+    {
+      "epoch": 112.28,
+      "learning_rate": 3.233944954128441e-05,
+      "loss": 0.1187,
+      "step": 12239
+    },
+    {
+      "epoch": 112.29,
+      "learning_rate": 3.230122324159022e-05,
+      "loss": 0.1086,
+      "step": 12240
+    },
+    {
+      "epoch": 112.3,
+      "learning_rate": 3.2262996941896024e-05,
+      "loss": 0.1248,
+      "step": 12241
+    },
+    {
+      "epoch": 112.31,
+      "learning_rate": 3.2224770642201835e-05,
+      "loss": 0.1143,
+      "step": 12242
+    },
+    {
+      "epoch": 112.32,
+      "learning_rate": 3.2186544342507647e-05,
+      "loss": 0.061,
+      "step": 12243
+    },
+    {
+      "epoch": 112.33,
+      "learning_rate": 3.214831804281346e-05,
+      "loss": 0.1367,
+      "step": 12244
+    },
+    {
+      "epoch": 112.34,
+      "learning_rate": 3.211009174311927e-05,
+      "loss": 0.0668,
+      "step": 12245
+    },
+    {
+      "epoch": 112.35,
+      "learning_rate": 3.207186544342508e-05,
+      "loss": 0.0869,
+      "step": 12246
+    },
+    {
+      "epoch": 112.36,
+      "learning_rate": 3.2033639143730884e-05,
+      "loss": 0.1234,
+      "step": 12247
+    },
+    {
+      "epoch": 112.37,
+      "learning_rate": 3.1995412844036695e-05,
+      "loss": 0.0902,
+      "step": 12248
+    },
+    {
+      "epoch": 112.38,
+      "learning_rate": 3.1957186544342506e-05,
+      "loss": 0.1014,
+      "step": 12249
+    },
+    {
+      "epoch": 112.39,
+      "learning_rate": 3.191896024464832e-05,
+      "loss": 0.0855,
+      "step": 12250
+    },
+    {
+      "epoch": 112.39,
+      "learning_rate": 3.188073394495413e-05,
+      "loss": 0.0454,
+      "step": 12251
+    },
+    {
+      "epoch": 112.4,
+      "learning_rate": 3.184250764525994e-05,
+      "loss": 0.0491,
+      "step": 12252
+    },
+    {
+      "epoch": 112.41,
+      "learning_rate": 3.180428134556575e-05,
+      "loss": 0.0874,
+      "step": 12253
+    },
+    {
+      "epoch": 112.42,
+      "learning_rate": 3.176605504587156e-05,
+      "loss": 0.0994,
+      "step": 12254
+    },
+    {
+      "epoch": 112.43,
+      "learning_rate": 3.172782874617737e-05,
+      "loss": 0.0227,
+      "step": 12255
+    },
+    {
+      "epoch": 112.44,
+      "learning_rate": 3.1689602446483184e-05,
+      "loss": 0.1269,
+      "step": 12256
+    },
+    {
+      "epoch": 112.45,
+      "learning_rate": 3.1651376146788995e-05,
+      "loss": 0.1329,
+      "step": 12257
+    },
+    {
+      "epoch": 112.46,
+      "learning_rate": 3.1613149847094806e-05,
+      "loss": 0.0819,
+      "step": 12258
+    },
+    {
+      "epoch": 112.47,
+      "learning_rate": 3.157492354740061e-05,
+      "loss": 0.0303,
+      "step": 12259
+    },
+    {
+      "epoch": 112.48,
+      "learning_rate": 3.153669724770642e-05,
+      "loss": 0.0774,
+      "step": 12260
+    },
+    {
+      "epoch": 112.49,
+      "learning_rate": 3.149847094801223e-05,
+      "loss": 0.0581,
+      "step": 12261
+    },
+    {
+      "epoch": 112.5,
+      "learning_rate": 3.146024464831804e-05,
+      "loss": 0.0145,
+      "step": 12262
+    },
+    {
+      "epoch": 112.5,
+      "learning_rate": 3.1422018348623854e-05,
+      "loss": 0.1462,
+      "step": 12263
+    },
+    {
+      "epoch": 112.51,
+      "learning_rate": 3.1383792048929665e-05,
+      "loss": 0.146,
+      "step": 12264
+    },
+    {
+      "epoch": 112.52,
+      "learning_rate": 3.134556574923547e-05,
+      "loss": 0.1195,
+      "step": 12265
+    },
+    {
+      "epoch": 112.53,
+      "learning_rate": 3.130733944954128e-05,
+      "loss": 0.1135,
+      "step": 12266
+    },
+    {
+      "epoch": 112.54,
+      "learning_rate": 3.126911314984709e-05,
+      "loss": 0.0902,
+      "step": 12267
+    },
+    {
+      "epoch": 112.55,
+      "learning_rate": 3.123088685015291e-05,
+      "loss": 0.0874,
+      "step": 12268
+    },
+    {
+      "epoch": 112.56,
+      "learning_rate": 3.119266055045872e-05,
+      "loss": 0.1366,
+      "step": 12269
+    },
+    {
+      "epoch": 112.57,
+      "learning_rate": 3.1154434250764525e-05,
+      "loss": 0.095,
+      "step": 12270
+    },
+    {
+      "epoch": 112.58,
+      "learning_rate": 3.1116207951070336e-05,
+      "loss": 0.1011,
+      "step": 12271
+    },
+    {
+      "epoch": 112.59,
+      "learning_rate": 3.107798165137615e-05,
+      "loss": 0.0484,
+      "step": 12272
+    },
+    {
+      "epoch": 112.6,
+      "learning_rate": 3.103975535168196e-05,
+      "loss": 0.1127,
+      "step": 12273
+    },
+    {
+      "epoch": 112.61,
+      "learning_rate": 3.100152905198777e-05,
+      "loss": 0.0875,
+      "step": 12274
+    },
+    {
+      "epoch": 112.61,
+      "learning_rate": 3.096330275229358e-05,
+      "loss": 0.0886,
+      "step": 12275
+    },
+    {
+      "epoch": 112.62,
+      "learning_rate": 3.092507645259939e-05,
+      "loss": 0.1165,
+      "step": 12276
+    },
+    {
+      "epoch": 112.63,
+      "learning_rate": 3.08868501529052e-05,
+      "loss": 0.0635,
+      "step": 12277
+    },
+    {
+      "epoch": 112.64,
+      "learning_rate": 3.0848623853211013e-05,
+      "loss": 0.0998,
+      "step": 12278
+    },
+    {
+      "epoch": 112.65,
+      "learning_rate": 3.081039755351682e-05,
+      "loss": 0.05,
+      "step": 12279
+    },
+    {
+      "epoch": 112.66,
+      "learning_rate": 3.077217125382263e-05,
+      "loss": 0.0469,
+      "step": 12280
+    },
+    {
+      "epoch": 112.67,
+      "learning_rate": 3.073394495412844e-05,
+      "loss": 0.0676,
+      "step": 12281
+    },
+    {
+      "epoch": 112.68,
+      "learning_rate": 3.069571865443425e-05,
+      "loss": 0.1004,
+      "step": 12282
+    },
+    {
+      "epoch": 112.69,
+      "learning_rate": 3.065749235474006e-05,
+      "loss": 0.0953,
+      "step": 12283
+    },
+    {
+      "epoch": 112.7,
+      "learning_rate": 3.061926605504587e-05,
+      "loss": 0.0945,
+      "step": 12284
+    },
+    {
+      "epoch": 112.71,
+      "learning_rate": 3.0581039755351684e-05,
+      "loss": 0.055,
+      "step": 12285
+    },
+    {
+      "epoch": 112.72,
+      "learning_rate": 3.0542813455657495e-05,
+      "loss": 0.1105,
+      "step": 12286
+    },
+    {
+      "epoch": 112.72,
+      "learning_rate": 3.0504587155963306e-05,
+      "loss": 0.1198,
+      "step": 12287
+    },
+    {
+      "epoch": 112.73,
+      "learning_rate": 3.0466360856269114e-05,
+      "loss": 0.1223,
+      "step": 12288
+    },
+    {
+      "epoch": 112.74,
+      "learning_rate": 3.0428134556574925e-05,
+      "loss": 0.1014,
+      "step": 12289
+    },
+    {
+      "epoch": 112.75,
+      "learning_rate": 3.0389908256880736e-05,
+      "loss": 0.118,
+      "step": 12290
+    },
+    {
+      "epoch": 112.76,
+      "learning_rate": 3.0351681957186544e-05,
+      "loss": 0.1285,
+      "step": 12291
+    },
+    {
+      "epoch": 112.77,
+      "learning_rate": 3.0313455657492355e-05,
+      "loss": 0.0933,
+      "step": 12292
+    },
+    {
+      "epoch": 112.78,
+      "learning_rate": 3.0275229357798166e-05,
+      "loss": 0.1402,
+      "step": 12293
+    },
+    {
+      "epoch": 112.79,
+      "learning_rate": 3.0237003058103977e-05,
+      "loss": 0.1542,
+      "step": 12294
+    },
+    {
+      "epoch": 112.8,
+      "learning_rate": 3.0198776758409788e-05,
+      "loss": 0.0928,
+      "step": 12295
+    },
+    {
+      "epoch": 112.81,
+      "learning_rate": 3.01605504587156e-05,
+      "loss": 0.1165,
+      "step": 12296
+    },
+    {
+      "epoch": 112.82,
+      "learning_rate": 3.0122324159021407e-05,
+      "loss": 0.0688,
+      "step": 12297
+    },
+    {
+      "epoch": 112.83,
+      "learning_rate": 3.0084097859327218e-05,
+      "loss": 0.0966,
+      "step": 12298
+    },
+    {
+      "epoch": 112.83,
+      "learning_rate": 3.004587155963303e-05,
+      "loss": 0.1053,
+      "step": 12299
+    },
+    {
+      "epoch": 112.84,
+      "learning_rate": 3.0007645259938837e-05,
+      "loss": 0.1018,
+      "step": 12300
+    },
+    {
+      "epoch": 112.85,
+      "learning_rate": 2.996941896024465e-05,
+      "loss": 0.1175,
+      "step": 12301
+    },
+    {
+      "epoch": 112.86,
+      "learning_rate": 2.9931192660550462e-05,
+      "loss": 0.0706,
+      "step": 12302
+    },
+    {
+      "epoch": 112.87,
+      "learning_rate": 2.989296636085627e-05,
+      "loss": 0.0987,
+      "step": 12303
+    },
+    {
+      "epoch": 112.88,
+      "learning_rate": 2.985474006116208e-05,
+      "loss": 0.1407,
+      "step": 12304
+    },
+    {
+      "epoch": 112.89,
+      "learning_rate": 2.9816513761467892e-05,
+      "loss": 0.0907,
+      "step": 12305
+    },
+    {
+      "epoch": 112.9,
+      "learning_rate": 2.97782874617737e-05,
+      "loss": 0.0806,
+      "step": 12306
+    },
+    {
+      "epoch": 112.91,
+      "learning_rate": 2.974006116207951e-05,
+      "loss": 0.0692,
+      "step": 12307
+    },
+    {
+      "epoch": 112.92,
+      "learning_rate": 2.9701834862385325e-05,
+      "loss": 0.1123,
+      "step": 12308
+    },
+    {
+      "epoch": 112.93,
+      "learning_rate": 2.9663608562691133e-05,
+      "loss": 0.068,
+      "step": 12309
+    },
+    {
+      "epoch": 112.94,
+      "learning_rate": 2.9625382262996944e-05,
+      "loss": 0.0642,
+      "step": 12310
+    },
+    {
+      "epoch": 112.94,
+      "learning_rate": 2.9587155963302755e-05,
+      "loss": 0.1191,
+      "step": 12311
+    },
+    {
+      "epoch": 112.95,
+      "learning_rate": 2.9548929663608563e-05,
+      "loss": 0.0596,
+      "step": 12312
+    },
+    {
+      "epoch": 112.96,
+      "learning_rate": 2.9510703363914374e-05,
+      "loss": 0.0572,
+      "step": 12313
+    },
+    {
+      "epoch": 112.97,
+      "learning_rate": 2.9472477064220185e-05,
+      "loss": 0.0741,
+      "step": 12314
+    },
+    {
+      "epoch": 112.98,
+      "learning_rate": 2.9434250764525992e-05,
+      "loss": 0.0696,
+      "step": 12315
+    },
+    {
+      "epoch": 112.99,
+      "learning_rate": 2.9396024464831807e-05,
+      "loss": 0.0574,
+      "step": 12316
+    },
+    {
+      "epoch": 113.0,
+      "learning_rate": 2.9357798165137618e-05,
+      "loss": 0.1027,
+      "step": 12317
+    },
+    {
+      "epoch": 113.01,
+      "learning_rate": 2.9319571865443426e-05,
+      "loss": 0.1564,
+      "step": 12318
+    },
+    {
+      "epoch": 113.02,
+      "learning_rate": 2.9281345565749237e-05,
+      "loss": 0.1126,
+      "step": 12319
+    },
+    {
+      "epoch": 113.03,
+      "learning_rate": 2.9243119266055048e-05,
+      "loss": 0.1184,
+      "step": 12320
+    },
+    {
+      "epoch": 113.04,
+      "learning_rate": 2.9204892966360855e-05,
+      "loss": 0.0933,
+      "step": 12321
+    },
+    {
+      "epoch": 113.05,
+      "learning_rate": 2.9166666666666666e-05,
+      "loss": 0.1026,
+      "step": 12322
+    },
+    {
+      "epoch": 113.06,
+      "learning_rate": 2.912844036697248e-05,
+      "loss": 0.0951,
+      "step": 12323
+    },
+    {
+      "epoch": 113.06,
+      "learning_rate": 2.909021406727829e-05,
+      "loss": 0.115,
+      "step": 12324
+    },
+    {
+      "epoch": 113.07,
+      "learning_rate": 2.90519877675841e-05,
+      "loss": 0.0984,
+      "step": 12325
+    },
+    {
+      "epoch": 113.08,
+      "learning_rate": 2.901376146788991e-05,
+      "loss": 0.0809,
+      "step": 12326
+    },
+    {
+      "epoch": 113.09,
+      "learning_rate": 2.8975535168195718e-05,
+      "loss": 0.06,
+      "step": 12327
+    },
+    {
+      "epoch": 113.1,
+      "learning_rate": 2.893730886850153e-05,
+      "loss": 0.0292,
+      "step": 12328
+    },
+    {
+      "epoch": 113.11,
+      "learning_rate": 2.889908256880734e-05,
+      "loss": 0.0737,
+      "step": 12329
+    },
+    {
+      "epoch": 113.12,
+      "learning_rate": 2.8860856269113148e-05,
+      "loss": 0.0383,
+      "step": 12330
+    },
+    {
+      "epoch": 113.13,
+      "learning_rate": 2.8822629969418963e-05,
+      "loss": 0.0873,
+      "step": 12331
+    },
+    {
+      "epoch": 113.14,
+      "learning_rate": 2.8784403669724774e-05,
+      "loss": 0.0812,
+      "step": 12332
+    },
+    {
+      "epoch": 113.15,
+      "learning_rate": 2.874617737003058e-05,
+      "loss": 0.0912,
+      "step": 12333
+    },
+    {
+      "epoch": 113.16,
+      "learning_rate": 2.8707951070336392e-05,
+      "loss": 0.1045,
+      "step": 12334
+    },
+    {
+      "epoch": 113.17,
+      "learning_rate": 2.8669724770642203e-05,
+      "loss": 0.0829,
+      "step": 12335
+    },
+    {
+      "epoch": 113.17,
+      "learning_rate": 2.863149847094801e-05,
+      "loss": 0.1101,
+      "step": 12336
+    },
+    {
+      "epoch": 113.18,
+      "learning_rate": 2.8593272171253822e-05,
+      "loss": 0.0754,
+      "step": 12337
+    },
+    {
+      "epoch": 113.19,
+      "learning_rate": 2.8555045871559637e-05,
+      "loss": 0.0382,
+      "step": 12338
+    },
+    {
+      "epoch": 113.2,
+      "learning_rate": 2.8516819571865444e-05,
+      "loss": 0.1004,
+      "step": 12339
+    },
+    {
+      "epoch": 113.21,
+      "learning_rate": 2.8478593272171255e-05,
+      "loss": 0.1214,
+      "step": 12340
+    },
+    {
+      "epoch": 113.22,
+      "learning_rate": 2.8440366972477066e-05,
+      "loss": 0.0773,
+      "step": 12341
+    },
+    {
+      "epoch": 113.23,
+      "learning_rate": 2.8402140672782874e-05,
+      "loss": 0.0966,
+      "step": 12342
+    },
+    {
+      "epoch": 113.24,
+      "learning_rate": 2.8363914373088685e-05,
+      "loss": 0.0602,
+      "step": 12343
+    },
+    {
+      "epoch": 113.25,
+      "learning_rate": 2.8325688073394496e-05,
+      "loss": 0.0553,
+      "step": 12344
+    },
+    {
+      "epoch": 113.26,
+      "learning_rate": 2.8287461773700307e-05,
+      "loss": 0.1936,
+      "step": 12345
+    },
+    {
+      "epoch": 113.27,
+      "learning_rate": 2.824923547400612e-05,
+      "loss": 0.1227,
+      "step": 12346
+    },
+    {
+      "epoch": 113.28,
+      "learning_rate": 2.821100917431193e-05,
+      "loss": 0.0971,
+      "step": 12347
+    },
+    {
+      "epoch": 113.28,
+      "learning_rate": 2.8172782874617737e-05,
+      "loss": 0.1105,
+      "step": 12348
+    },
+    {
+      "epoch": 113.29,
+      "learning_rate": 2.8134556574923548e-05,
+      "loss": 0.0821,
+      "step": 12349
+    },
+    {
+      "epoch": 113.3,
+      "learning_rate": 2.809633027522936e-05,
+      "loss": 0.1016,
+      "step": 12350
+    },
+    {
+      "epoch": 113.31,
+      "learning_rate": 2.8058103975535167e-05,
+      "loss": 0.046,
+      "step": 12351
+    },
+    {
+      "epoch": 113.32,
+      "learning_rate": 2.8019877675840978e-05,
+      "loss": 0.0928,
+      "step": 12352
+    },
+    {
+      "epoch": 113.33,
+      "learning_rate": 2.7981651376146792e-05,
+      "loss": 0.0771,
+      "step": 12353
+    },
+    {
+      "epoch": 113.34,
+      "learning_rate": 2.79434250764526e-05,
+      "loss": 0.0686,
+      "step": 12354
+    },
+    {
+      "epoch": 113.35,
+      "learning_rate": 2.790519877675841e-05,
+      "loss": 0.0626,
+      "step": 12355
+    },
+    {
+      "epoch": 113.36,
+      "learning_rate": 2.7866972477064222e-05,
+      "loss": 0.1199,
+      "step": 12356
+    },
+    {
+      "epoch": 113.37,
+      "learning_rate": 2.782874617737003e-05,
+      "loss": 0.1065,
+      "step": 12357
+    },
+    {
+      "epoch": 113.38,
+      "learning_rate": 2.779051987767584e-05,
+      "loss": 0.1277,
+      "step": 12358
+    },
+    {
+      "epoch": 113.39,
+      "learning_rate": 2.7752293577981652e-05,
+      "loss": 0.1286,
+      "step": 12359
+    },
+    {
+      "epoch": 113.39,
+      "learning_rate": 2.7714067278287463e-05,
+      "loss": 0.0681,
+      "step": 12360
+    },
+    {
+      "epoch": 113.4,
+      "learning_rate": 2.7675840978593274e-05,
+      "loss": 0.0562,
+      "step": 12361
+    },
+    {
+      "epoch": 113.41,
+      "learning_rate": 2.7637614678899085e-05,
+      "loss": 0.0906,
+      "step": 12362
+    },
+    {
+      "epoch": 113.42,
+      "learning_rate": 2.7599388379204893e-05,
+      "loss": 0.0795,
+      "step": 12363
+    },
+    {
+      "epoch": 113.43,
+      "learning_rate": 2.7561162079510704e-05,
+      "loss": 0.0977,
+      "step": 12364
+    },
+    {
+      "epoch": 113.44,
+      "learning_rate": 2.7522935779816515e-05,
+      "loss": 0.0821,
+      "step": 12365
+    },
+    {
+      "epoch": 113.45,
+      "learning_rate": 2.7484709480122323e-05,
+      "loss": 0.1382,
+      "step": 12366
+    },
+    {
+      "epoch": 113.46,
+      "learning_rate": 2.7446483180428137e-05,
+      "loss": 0.0782,
+      "step": 12367
+    },
+    {
+      "epoch": 113.47,
+      "learning_rate": 2.7408256880733948e-05,
+      "loss": 0.0487,
+      "step": 12368
+    },
+    {
+      "epoch": 113.48,
+      "learning_rate": 2.7370030581039756e-05,
+      "loss": 0.1235,
+      "step": 12369
+    },
+    {
+      "epoch": 113.49,
+      "learning_rate": 2.7331804281345567e-05,
+      "loss": 0.0954,
+      "step": 12370
+    },
+    {
+      "epoch": 113.5,
+      "learning_rate": 2.7293577981651378e-05,
+      "loss": 0.0476,
+      "step": 12371
+    },
+    {
+      "epoch": 113.5,
+      "learning_rate": 2.7255351681957186e-05,
+      "loss": 0.161,
+      "step": 12372
+    },
+    {
+      "epoch": 113.51,
+      "learning_rate": 2.7217125382262997e-05,
+      "loss": 0.1715,
+      "step": 12373
+    },
+    {
+      "epoch": 113.52,
+      "learning_rate": 2.7178899082568808e-05,
+      "loss": 0.0871,
+      "step": 12374
+    },
+    {
+      "epoch": 113.53,
+      "learning_rate": 2.714067278287462e-05,
+      "loss": 0.0891,
+      "step": 12375
+    },
+    {
+      "epoch": 113.54,
+      "learning_rate": 2.710244648318043e-05,
+      "loss": 0.1089,
+      "step": 12376
+    },
+    {
+      "epoch": 113.55,
+      "learning_rate": 2.706422018348624e-05,
+      "loss": 0.1032,
+      "step": 12377
+    },
+    {
+      "epoch": 113.56,
+      "learning_rate": 2.702599388379205e-05,
+      "loss": 0.0999,
+      "step": 12378
+    },
+    {
+      "epoch": 113.57,
+      "learning_rate": 2.698776758409786e-05,
+      "loss": 0.0757,
+      "step": 12379
+    },
+    {
+      "epoch": 113.58,
+      "learning_rate": 2.694954128440367e-05,
+      "loss": 0.077,
+      "step": 12380
+    },
+    {
+      "epoch": 113.59,
+      "learning_rate": 2.691131498470948e-05,
+      "loss": 0.1035,
+      "step": 12381
+    },
+    {
+      "epoch": 113.6,
+      "learning_rate": 2.6873088685015293e-05,
+      "loss": 0.1509,
+      "step": 12382
+    },
+    {
+      "epoch": 113.61,
+      "learning_rate": 2.6834862385321104e-05,
+      "loss": 0.0677,
+      "step": 12383
+    },
+    {
+      "epoch": 113.61,
+      "learning_rate": 2.679663608562691e-05,
+      "loss": 0.0636,
+      "step": 12384
+    },
+    {
+      "epoch": 113.62,
+      "learning_rate": 2.6758409785932723e-05,
+      "loss": 0.0783,
+      "step": 12385
+    },
+    {
+      "epoch": 113.63,
+      "learning_rate": 2.6720183486238534e-05,
+      "loss": 0.0626,
+      "step": 12386
+    },
+    {
+      "epoch": 113.64,
+      "learning_rate": 2.668195718654434e-05,
+      "loss": 0.0303,
+      "step": 12387
+    },
+    {
+      "epoch": 113.65,
+      "learning_rate": 2.6643730886850153e-05,
+      "loss": 0.1088,
+      "step": 12388
+    },
+    {
+      "epoch": 113.66,
+      "learning_rate": 2.6605504587155967e-05,
+      "loss": 0.0665,
+      "step": 12389
+    },
+    {
+      "epoch": 113.67,
+      "learning_rate": 2.6567278287461775e-05,
+      "loss": 0.0997,
+      "step": 12390
+    },
+    {
+      "epoch": 113.68,
+      "learning_rate": 2.6529051987767586e-05,
+      "loss": 0.0731,
+      "step": 12391
+    },
+    {
+      "epoch": 113.69,
+      "learning_rate": 2.6490825688073397e-05,
+      "loss": 0.0757,
+      "step": 12392
+    },
+    {
+      "epoch": 113.7,
+      "learning_rate": 2.6452599388379204e-05,
+      "loss": 0.0259,
+      "step": 12393
+    },
+    {
+      "epoch": 113.71,
+      "learning_rate": 2.6414373088685016e-05,
+      "loss": 0.1105,
+      "step": 12394
+    },
+    {
+      "epoch": 113.72,
+      "learning_rate": 2.6376146788990827e-05,
+      "loss": 0.0811,
+      "step": 12395
+    },
+    {
+      "epoch": 113.72,
+      "learning_rate": 2.6337920489296634e-05,
+      "loss": 0.056,
+      "step": 12396
+    },
+    {
+      "epoch": 113.73,
+      "learning_rate": 2.629969418960245e-05,
+      "loss": 0.0149,
+      "step": 12397
+    },
+    {
+      "epoch": 113.74,
+      "learning_rate": 2.626146788990826e-05,
+      "loss": 0.0135,
+      "step": 12398
+    },
+    {
+      "epoch": 113.75,
+      "learning_rate": 2.6223241590214067e-05,
+      "loss": 0.196,
+      "step": 12399
+    },
+    {
+      "epoch": 113.76,
+      "learning_rate": 2.618501529051988e-05,
+      "loss": 0.1218,
+      "step": 12400
+    },
+    {
+      "epoch": 113.77,
+      "learning_rate": 2.614678899082569e-05,
+      "loss": 0.0653,
+      "step": 12401
+    },
+    {
+      "epoch": 113.78,
+      "learning_rate": 2.6108562691131497e-05,
+      "loss": 0.0884,
+      "step": 12402
+    },
+    {
+      "epoch": 113.79,
+      "learning_rate": 2.607033639143731e-05,
+      "loss": 0.0659,
+      "step": 12403
+    },
+    {
+      "epoch": 113.8,
+      "learning_rate": 2.6032110091743123e-05,
+      "loss": 0.0912,
+      "step": 12404
+    },
+    {
+      "epoch": 113.81,
+      "learning_rate": 2.599388379204893e-05,
+      "loss": 0.1029,
+      "step": 12405
+    },
+    {
+      "epoch": 113.82,
+      "learning_rate": 2.595565749235474e-05,
+      "loss": 0.0884,
+      "step": 12406
+    },
+    {
+      "epoch": 113.83,
+      "learning_rate": 2.5917431192660553e-05,
+      "loss": 0.1036,
+      "step": 12407
+    },
+    {
+      "epoch": 113.83,
+      "learning_rate": 2.587920489296636e-05,
+      "loss": 0.1305,
+      "step": 12408
+    },
+    {
+      "epoch": 113.84,
+      "learning_rate": 2.584097859327217e-05,
+      "loss": 0.0581,
+      "step": 12409
+    },
+    {
+      "epoch": 113.85,
+      "learning_rate": 2.5802752293577982e-05,
+      "loss": 0.1085,
+      "step": 12410
+    },
+    {
+      "epoch": 113.86,
+      "learning_rate": 2.576452599388379e-05,
+      "loss": 0.0594,
+      "step": 12411
+    },
+    {
+      "epoch": 113.87,
+      "learning_rate": 2.5726299694189605e-05,
+      "loss": 0.0594,
+      "step": 12412
+    },
+    {
+      "epoch": 113.88,
+      "learning_rate": 2.5688073394495416e-05,
+      "loss": 0.115,
+      "step": 12413
+    },
+    {
+      "epoch": 113.89,
+      "learning_rate": 2.5649847094801223e-05,
+      "loss": 0.0596,
+      "step": 12414
+    },
+    {
+      "epoch": 113.9,
+      "learning_rate": 2.5611620795107034e-05,
+      "loss": 0.0754,
+      "step": 12415
+    },
+    {
+      "epoch": 113.91,
+      "learning_rate": 2.5573394495412845e-05,
+      "loss": 0.1323,
+      "step": 12416
+    },
+    {
+      "epoch": 113.92,
+      "learning_rate": 2.5535168195718653e-05,
+      "loss": 0.1084,
+      "step": 12417
+    },
+    {
+      "epoch": 113.93,
+      "learning_rate": 2.5496941896024464e-05,
+      "loss": 0.0765,
+      "step": 12418
+    },
+    {
+      "epoch": 113.94,
+      "learning_rate": 2.545871559633028e-05,
+      "loss": 0.0452,
+      "step": 12419
+    },
+    {
+      "epoch": 113.94,
+      "learning_rate": 2.5420489296636086e-05,
+      "loss": 0.0905,
+      "step": 12420
+    },
+    {
+      "epoch": 113.95,
+      "learning_rate": 2.5382262996941897e-05,
+      "loss": 0.0749,
+      "step": 12421
+    },
+    {
+      "epoch": 113.96,
+      "learning_rate": 2.534403669724771e-05,
+      "loss": 0.061,
+      "step": 12422
+    },
+    {
+      "epoch": 113.97,
+      "learning_rate": 2.5305810397553516e-05,
+      "loss": 0.0616,
+      "step": 12423
+    },
+    {
+      "epoch": 113.98,
+      "learning_rate": 2.5267584097859327e-05,
+      "loss": 0.0942,
+      "step": 12424
+    },
+    {
+      "epoch": 113.99,
+      "learning_rate": 2.5229357798165138e-05,
+      "loss": 0.0364,
+      "step": 12425
+    },
+    {
+      "epoch": 114.0,
+      "learning_rate": 2.519113149847095e-05,
+      "loss": 0.0729,
+      "step": 12426
+    },
+    {
+      "epoch": 114.01,
+      "learning_rate": 2.515290519877676e-05,
+      "loss": 0.1266,
+      "step": 12427
+    },
+    {
+      "epoch": 114.02,
+      "learning_rate": 2.511467889908257e-05,
+      "loss": 0.0778,
+      "step": 12428
+    },
+    {
+      "epoch": 114.03,
+      "learning_rate": 2.507645259938838e-05,
+      "loss": 0.1095,
+      "step": 12429
+    },
+    {
+      "epoch": 114.04,
+      "learning_rate": 2.503822629969419e-05,
+      "loss": 0.1187,
+      "step": 12430
+    },
+    {
+      "epoch": 114.05,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0845,
+      "step": 12431
+    },
+    {
+      "epoch": 114.06,
+      "learning_rate": 2.496177370030581e-05,
+      "loss": 0.0905,
+      "step": 12432
+    },
+    {
+      "epoch": 114.06,
+      "learning_rate": 2.492354740061162e-05,
+      "loss": 0.098,
+      "step": 12433
+    },
+    {
+      "epoch": 114.07,
+      "learning_rate": 2.4885321100917434e-05,
+      "loss": 0.0753,
+      "step": 12434
+    },
+    {
+      "epoch": 114.08,
+      "learning_rate": 2.4847094801223242e-05,
+      "loss": 0.0588,
+      "step": 12435
+    },
+    {
+      "epoch": 114.09,
+      "learning_rate": 2.4808868501529053e-05,
+      "loss": 0.0529,
+      "step": 12436
+    },
+    {
+      "epoch": 114.1,
+      "learning_rate": 2.4770642201834864e-05,
+      "loss": 0.1098,
+      "step": 12437
+    },
+    {
+      "epoch": 114.11,
+      "learning_rate": 2.4732415902140672e-05,
+      "loss": 0.1287,
+      "step": 12438
+    },
+    {
+      "epoch": 114.12,
+      "learning_rate": 2.4694189602446483e-05,
+      "loss": 0.0953,
+      "step": 12439
+    },
+    {
+      "epoch": 114.13,
+      "learning_rate": 2.4655963302752294e-05,
+      "loss": 0.0831,
+      "step": 12440
+    },
+    {
+      "epoch": 114.14,
+      "learning_rate": 2.4617737003058105e-05,
+      "loss": 0.1038,
+      "step": 12441
+    },
+    {
+      "epoch": 114.15,
+      "learning_rate": 2.4579510703363916e-05,
+      "loss": 0.0418,
+      "step": 12442
+    },
+    {
+      "epoch": 114.16,
+      "learning_rate": 2.4541284403669727e-05,
+      "loss": 0.063,
+      "step": 12443
+    },
+    {
+      "epoch": 114.17,
+      "learning_rate": 2.4503058103975535e-05,
+      "loss": 0.064,
+      "step": 12444
+    },
+    {
+      "epoch": 114.17,
+      "learning_rate": 2.4464831804281346e-05,
+      "loss": 0.0747,
+      "step": 12445
+    },
+    {
+      "epoch": 114.18,
+      "learning_rate": 2.4426605504587157e-05,
+      "loss": 0.0493,
+      "step": 12446
+    },
+    {
+      "epoch": 114.19,
+      "learning_rate": 2.4388379204892965e-05,
+      "loss": 0.0934,
+      "step": 12447
+    },
+    {
+      "epoch": 114.2,
+      "learning_rate": 2.435015290519878e-05,
+      "loss": 0.0801,
+      "step": 12448
+    },
+    {
+      "epoch": 114.21,
+      "learning_rate": 2.431192660550459e-05,
+      "loss": 0.0798,
+      "step": 12449
+    },
+    {
+      "epoch": 114.22,
+      "learning_rate": 2.4273700305810398e-05,
+      "loss": 0.0473,
+      "step": 12450
+    },
+    {
+      "epoch": 114.23,
+      "learning_rate": 2.423547400611621e-05,
+      "loss": 0.0214,
+      "step": 12451
+    },
+    {
+      "epoch": 114.24,
+      "learning_rate": 2.419724770642202e-05,
+      "loss": 0.0475,
+      "step": 12452
+    },
+    {
+      "epoch": 114.25,
+      "learning_rate": 2.4159021406727828e-05,
+      "loss": 0.0397,
+      "step": 12453
+    },
+    {
+      "epoch": 114.26,
+      "learning_rate": 2.412079510703364e-05,
+      "loss": 0.1523,
+      "step": 12454
+    },
+    {
+      "epoch": 114.27,
+      "learning_rate": 2.408256880733945e-05,
+      "loss": 0.1307,
+      "step": 12455
+    },
+    {
+      "epoch": 114.28,
+      "learning_rate": 2.404434250764526e-05,
+      "loss": 0.0995,
+      "step": 12456
+    },
+    {
+      "epoch": 114.28,
+      "learning_rate": 2.4006116207951072e-05,
+      "loss": 0.1434,
+      "step": 12457
+    },
+    {
+      "epoch": 114.29,
+      "learning_rate": 2.3967889908256883e-05,
+      "loss": 0.1048,
+      "step": 12458
+    },
+    {
+      "epoch": 114.3,
+      "learning_rate": 2.392966360856269e-05,
+      "loss": 0.1312,
+      "step": 12459
+    },
+    {
+      "epoch": 114.31,
+      "learning_rate": 2.3891437308868502e-05,
+      "loss": 0.0742,
+      "step": 12460
+    },
+    {
+      "epoch": 114.32,
+      "learning_rate": 2.3853211009174313e-05,
+      "loss": 0.0955,
+      "step": 12461
+    },
+    {
+      "epoch": 114.33,
+      "learning_rate": 2.381498470948012e-05,
+      "loss": 0.0398,
+      "step": 12462
+    },
+    {
+      "epoch": 114.34,
+      "learning_rate": 2.3776758409785935e-05,
+      "loss": 0.0763,
+      "step": 12463
+    },
+    {
+      "epoch": 114.35,
+      "learning_rate": 2.3738532110091746e-05,
+      "loss": 0.1178,
+      "step": 12464
+    },
+    {
+      "epoch": 114.36,
+      "learning_rate": 2.3700305810397554e-05,
+      "loss": 0.0552,
+      "step": 12465
+    },
+    {
+      "epoch": 114.37,
+      "learning_rate": 2.3662079510703365e-05,
+      "loss": 0.102,
+      "step": 12466
+    },
+    {
+      "epoch": 114.38,
+      "learning_rate": 2.3623853211009176e-05,
+      "loss": 0.051,
+      "step": 12467
+    },
+    {
+      "epoch": 114.39,
+      "learning_rate": 2.3585626911314983e-05,
+      "loss": 0.0482,
+      "step": 12468
+    },
+    {
+      "epoch": 114.39,
+      "learning_rate": 2.3547400611620795e-05,
+      "loss": 0.0626,
+      "step": 12469
+    },
+    {
+      "epoch": 114.4,
+      "learning_rate": 2.350917431192661e-05,
+      "loss": 0.1343,
+      "step": 12470
+    },
+    {
+      "epoch": 114.41,
+      "learning_rate": 2.3470948012232417e-05,
+      "loss": 0.0656,
+      "step": 12471
+    },
+    {
+      "epoch": 114.42,
+      "learning_rate": 2.3432721712538228e-05,
+      "loss": 0.0696,
+      "step": 12472
+    },
+    {
+      "epoch": 114.43,
+      "learning_rate": 2.339449541284404e-05,
+      "loss": 0.1073,
+      "step": 12473
+    },
+    {
+      "epoch": 114.44,
+      "learning_rate": 2.3356269113149846e-05,
+      "loss": 0.0883,
+      "step": 12474
+    },
+    {
+      "epoch": 114.45,
+      "learning_rate": 2.3318042813455658e-05,
+      "loss": 0.0981,
+      "step": 12475
+    },
+    {
+      "epoch": 114.46,
+      "learning_rate": 2.327981651376147e-05,
+      "loss": 0.0443,
+      "step": 12476
+    },
+    {
+      "epoch": 114.47,
+      "learning_rate": 2.3241590214067276e-05,
+      "loss": 0.082,
+      "step": 12477
+    },
+    {
+      "epoch": 114.48,
+      "learning_rate": 2.320336391437309e-05,
+      "loss": 0.0667,
+      "step": 12478
+    },
+    {
+      "epoch": 114.49,
+      "learning_rate": 2.3165137614678902e-05,
+      "loss": 0.0471,
+      "step": 12479
+    },
+    {
+      "epoch": 114.5,
+      "learning_rate": 2.312691131498471e-05,
+      "loss": 0.0323,
+      "step": 12480
+    },
+    {
+      "epoch": 114.5,
+      "learning_rate": 2.308868501529052e-05,
+      "loss": 0.1452,
+      "step": 12481
+    },
+    {
+      "epoch": 114.51,
+      "learning_rate": 2.305045871559633e-05,
+      "loss": 0.1673,
+      "step": 12482
+    },
+    {
+      "epoch": 114.52,
+      "learning_rate": 2.301223241590214e-05,
+      "loss": 0.0848,
+      "step": 12483
+    },
+    {
+      "epoch": 114.53,
+      "learning_rate": 2.297400611620795e-05,
+      "loss": 0.0895,
+      "step": 12484
+    },
+    {
+      "epoch": 114.54,
+      "learning_rate": 2.2935779816513765e-05,
+      "loss": 0.1248,
+      "step": 12485
+    },
+    {
+      "epoch": 114.55,
+      "learning_rate": 2.2897553516819572e-05,
+      "loss": 0.1021,
+      "step": 12486
+    },
+    {
+      "epoch": 114.56,
+      "learning_rate": 2.2859327217125384e-05,
+      "loss": 0.1396,
+      "step": 12487
+    },
+    {
+      "epoch": 114.57,
+      "learning_rate": 2.2821100917431195e-05,
+      "loss": 0.0838,
+      "step": 12488
+    },
+    {
+      "epoch": 114.58,
+      "learning_rate": 2.2782874617737002e-05,
+      "loss": 0.0883,
+      "step": 12489
+    },
+    {
+      "epoch": 114.59,
+      "learning_rate": 2.2744648318042813e-05,
+      "loss": 0.0941,
+      "step": 12490
+    },
+    {
+      "epoch": 114.6,
+      "learning_rate": 2.2706422018348624e-05,
+      "loss": 0.0997,
+      "step": 12491
+    },
+    {
+      "epoch": 114.61,
+      "learning_rate": 2.2668195718654432e-05,
+      "loss": 0.0938,
+      "step": 12492
+    },
+    {
+      "epoch": 114.61,
+      "learning_rate": 2.2629969418960247e-05,
+      "loss": 0.0726,
+      "step": 12493
+    },
+    {
+      "epoch": 114.62,
+      "learning_rate": 2.2591743119266058e-05,
+      "loss": 0.0486,
+      "step": 12494
+    },
+    {
+      "epoch": 114.63,
+      "learning_rate": 2.2553516819571865e-05,
+      "loss": 0.0797,
+      "step": 12495
+    },
+    {
+      "epoch": 114.64,
+      "learning_rate": 2.2515290519877676e-05,
+      "loss": 0.0813,
+      "step": 12496
+    },
+    {
+      "epoch": 114.65,
+      "learning_rate": 2.2477064220183487e-05,
+      "loss": 0.1147,
+      "step": 12497
+    },
+    {
+      "epoch": 114.66,
+      "learning_rate": 2.2438837920489295e-05,
+      "loss": 0.0797,
+      "step": 12498
+    },
+    {
+      "epoch": 114.67,
+      "learning_rate": 2.2400611620795106e-05,
+      "loss": 0.1145,
+      "step": 12499
+    },
+    {
+      "epoch": 114.68,
+      "learning_rate": 2.236238532110092e-05,
+      "loss": 0.0437,
+      "step": 12500
+    },
+    {
+      "epoch": 114.68,
+      "eval_cer": 0.13283557474624808,
+      "eval_loss": 0.7524892091751099,
+      "eval_runtime": 87.7282,
+      "eval_samples_per_second": 18.774,
+      "eval_steps_per_second": 2.348,
+      "eval_wer": 0.4640996833826984,
+      "step": 12500
+    },
+    {
+      "epoch": 114.69,
+      "learning_rate": 2.2324159021406728e-05,
+      "loss": 0.0644,
+      "step": 12501
+    },
+    {
+      "epoch": 114.7,
+      "learning_rate": 2.228593272171254e-05,
+      "loss": 0.0688,
+      "step": 12502
+    },
+    {
+      "epoch": 114.71,
+      "learning_rate": 2.224770642201835e-05,
+      "loss": 0.0649,
+      "step": 12503
+    },
+    {
+      "epoch": 114.72,
+      "learning_rate": 2.2209480122324158e-05,
+      "loss": 0.0559,
+      "step": 12504
+    },
+    {
+      "epoch": 114.72,
+      "learning_rate": 2.217125382262997e-05,
+      "loss": 0.1261,
+      "step": 12505
+    },
+    {
+      "epoch": 114.73,
+      "learning_rate": 2.213302752293578e-05,
+      "loss": 0.0548,
+      "step": 12506
+    },
+    {
+      "epoch": 114.74,
+      "learning_rate": 2.209480122324159e-05,
+      "loss": 0.0299,
+      "step": 12507
+    },
+    {
+      "epoch": 114.75,
+      "learning_rate": 2.2056574923547402e-05,
+      "loss": 0.111,
+      "step": 12508
+    },
+    {
+      "epoch": 114.76,
+      "learning_rate": 2.2018348623853213e-05,
+      "loss": 0.0901,
+      "step": 12509
+    },
+    {
+      "epoch": 114.77,
+      "learning_rate": 2.198012232415902e-05,
+      "loss": 0.1016,
+      "step": 12510
+    },
+    {
+      "epoch": 114.78,
+      "learning_rate": 2.1941896024464832e-05,
+      "loss": 0.1067,
+      "step": 12511
+    },
+    {
+      "epoch": 114.79,
+      "learning_rate": 2.1903669724770643e-05,
+      "loss": 0.1232,
+      "step": 12512
+    },
+    {
+      "epoch": 114.8,
+      "learning_rate": 2.186544342507645e-05,
+      "loss": 0.1065,
+      "step": 12513
+    },
+    {
+      "epoch": 114.81,
+      "learning_rate": 2.1827217125382262e-05,
+      "loss": 0.1063,
+      "step": 12514
+    },
+    {
+      "epoch": 114.82,
+      "learning_rate": 2.1788990825688076e-05,
+      "loss": 0.1113,
+      "step": 12515
+    },
+    {
+      "epoch": 114.83,
+      "learning_rate": 2.1750764525993884e-05,
+      "loss": 0.1007,
+      "step": 12516
+    },
+    {
+      "epoch": 114.83,
+      "learning_rate": 2.1712538226299695e-05,
+      "loss": 0.0892,
+      "step": 12517
+    },
+    {
+      "epoch": 114.84,
+      "learning_rate": 2.1674311926605506e-05,
+      "loss": 0.1287,
+      "step": 12518
+    },
+    {
+      "epoch": 114.85,
+      "learning_rate": 2.1636085626911314e-05,
+      "loss": 0.0821,
+      "step": 12519
+    },
+    {
+      "epoch": 114.86,
+      "learning_rate": 2.1597859327217125e-05,
+      "loss": 0.0936,
+      "step": 12520
+    },
+    {
+      "epoch": 114.87,
+      "learning_rate": 2.1559633027522936e-05,
+      "loss": 0.0819,
+      "step": 12521
+    },
+    {
+      "epoch": 114.88,
+      "learning_rate": 2.1521406727828747e-05,
+      "loss": 0.1081,
+      "step": 12522
+    },
+    {
+      "epoch": 114.89,
+      "learning_rate": 2.1483180428134558e-05,
+      "loss": 0.139,
+      "step": 12523
+    },
+    {
+      "epoch": 114.9,
+      "learning_rate": 2.144495412844037e-05,
+      "loss": 0.078,
+      "step": 12524
+    },
+    {
+      "epoch": 114.91,
+      "learning_rate": 2.1406727828746177e-05,
+      "loss": 0.0901,
+      "step": 12525
+    },
+    {
+      "epoch": 114.92,
+      "learning_rate": 2.1368501529051988e-05,
+      "loss": 0.1087,
+      "step": 12526
+    },
+    {
+      "epoch": 114.93,
+      "learning_rate": 2.13302752293578e-05,
+      "loss": 0.088,
+      "step": 12527
+    },
+    {
+      "epoch": 114.94,
+      "learning_rate": 2.1292048929663607e-05,
+      "loss": 0.0555,
+      "step": 12528
+    },
+    {
+      "epoch": 114.94,
+      "learning_rate": 2.125382262996942e-05,
+      "loss": 0.0617,
+      "step": 12529
+    },
+    {
+      "epoch": 114.95,
+      "learning_rate": 2.1215596330275232e-05,
+      "loss": 0.0484,
+      "step": 12530
+    },
+    {
+      "epoch": 114.96,
+      "learning_rate": 2.117737003058104e-05,
+      "loss": 0.046,
+      "step": 12531
+    },
+    {
+      "epoch": 114.97,
+      "learning_rate": 2.113914373088685e-05,
+      "loss": 0.0822,
+      "step": 12532
+    },
+    {
+      "epoch": 114.98,
+      "learning_rate": 2.1100917431192662e-05,
+      "loss": 0.1153,
+      "step": 12533
+    },
+    {
+      "epoch": 114.99,
+      "learning_rate": 2.106269113149847e-05,
+      "loss": 0.0747,
+      "step": 12534
+    },
+    {
+      "epoch": 115.0,
+      "learning_rate": 2.102446483180428e-05,
+      "loss": 0.1246,
+      "step": 12535
+    },
+    {
+      "epoch": 115.01,
+      "learning_rate": 2.0986238532110092e-05,
+      "loss": 0.0917,
+      "step": 12536
+    },
+    {
+      "epoch": 115.02,
+      "learning_rate": 2.0948012232415903e-05,
+      "loss": 0.0765,
+      "step": 12537
+    },
+    {
+      "epoch": 115.03,
+      "learning_rate": 2.0909785932721714e-05,
+      "loss": 0.0678,
+      "step": 12538
+    },
+    {
+      "epoch": 115.04,
+      "learning_rate": 2.0871559633027525e-05,
+      "loss": 0.1246,
+      "step": 12539
+    },
+    {
+      "epoch": 115.05,
+      "learning_rate": 2.0833333333333333e-05,
+      "loss": 0.0906,
+      "step": 12540
+    },
+    {
+      "epoch": 115.06,
+      "learning_rate": 2.0795107033639144e-05,
+      "loss": 0.0879,
+      "step": 12541
+    },
+    {
+      "epoch": 115.06,
+      "learning_rate": 2.0756880733944955e-05,
+      "loss": 0.1332,
+      "step": 12542
+    },
+    {
+      "epoch": 115.07,
+      "learning_rate": 2.0718654434250762e-05,
+      "loss": 0.0963,
+      "step": 12543
+    },
+    {
+      "epoch": 115.08,
+      "learning_rate": 2.0680428134556577e-05,
+      "loss": 0.101,
+      "step": 12544
+    },
+    {
+      "epoch": 115.09,
+      "learning_rate": 2.0642201834862388e-05,
+      "loss": 0.106,
+      "step": 12545
+    },
+    {
+      "epoch": 115.1,
+      "learning_rate": 2.0603975535168196e-05,
+      "loss": 0.0959,
+      "step": 12546
+    },
+    {
+      "epoch": 115.11,
+      "learning_rate": 2.0565749235474007e-05,
+      "loss": 0.1304,
+      "step": 12547
+    },
+    {
+      "epoch": 115.12,
+      "learning_rate": 2.0527522935779818e-05,
+      "loss": 0.1163,
+      "step": 12548
+    },
+    {
+      "epoch": 115.13,
+      "learning_rate": 2.0489296636085625e-05,
+      "loss": 0.0799,
+      "step": 12549
+    },
+    {
+      "epoch": 115.14,
+      "learning_rate": 2.0451070336391437e-05,
+      "loss": 0.0839,
+      "step": 12550
+    },
+    {
+      "epoch": 115.15,
+      "learning_rate": 2.041284403669725e-05,
+      "loss": 0.062,
+      "step": 12551
+    },
+    {
+      "epoch": 115.16,
+      "learning_rate": 2.037461773700306e-05,
+      "loss": 0.0704,
+      "step": 12552
+    },
+    {
+      "epoch": 115.17,
+      "learning_rate": 2.033639143730887e-05,
+      "loss": 0.0729,
+      "step": 12553
+    },
+    {
+      "epoch": 115.17,
+      "learning_rate": 2.029816513761468e-05,
+      "loss": 0.1296,
+      "step": 12554
+    },
+    {
+      "epoch": 115.18,
+      "learning_rate": 2.025993883792049e-05,
+      "loss": 0.126,
+      "step": 12555
+    },
+    {
+      "epoch": 115.19,
+      "learning_rate": 2.02217125382263e-05,
+      "loss": 0.0763,
+      "step": 12556
+    },
+    {
+      "epoch": 115.2,
+      "learning_rate": 2.018348623853211e-05,
+      "loss": 0.0657,
+      "step": 12557
+    },
+    {
+      "epoch": 115.21,
+      "learning_rate": 2.0145259938837918e-05,
+      "loss": 0.089,
+      "step": 12558
+    },
+    {
+      "epoch": 115.22,
+      "learning_rate": 2.0107033639143733e-05,
+      "loss": 0.1113,
+      "step": 12559
+    },
+    {
+      "epoch": 115.23,
+      "learning_rate": 2.0068807339449544e-05,
+      "loss": 0.055,
+      "step": 12560
+    },
+    {
+      "epoch": 115.24,
+      "learning_rate": 2.003058103975535e-05,
+      "loss": 0.0417,
+      "step": 12561
+    },
+    {
+      "epoch": 115.25,
+      "learning_rate": 1.9992354740061162e-05,
+      "loss": 0.0574,
+      "step": 12562
+    },
+    {
+      "epoch": 115.26,
+      "learning_rate": 1.9954128440366974e-05,
+      "loss": 0.0943,
+      "step": 12563
+    },
+    {
+      "epoch": 115.27,
+      "learning_rate": 1.991590214067278e-05,
+      "loss": 0.1018,
+      "step": 12564
+    },
+    {
+      "epoch": 115.28,
+      "learning_rate": 1.9877675840978592e-05,
+      "loss": 0.1116,
+      "step": 12565
+    },
+    {
+      "epoch": 115.28,
+      "learning_rate": 1.9839449541284407e-05,
+      "loss": 0.1213,
+      "step": 12566
+    },
+    {
+      "epoch": 115.29,
+      "learning_rate": 1.9801223241590214e-05,
+      "loss": 0.0716,
+      "step": 12567
+    },
+    {
+      "epoch": 115.3,
+      "learning_rate": 1.9762996941896025e-05,
+      "loss": 0.1072,
+      "step": 12568
+    },
+    {
+      "epoch": 115.31,
+      "learning_rate": 1.9724770642201837e-05,
+      "loss": 0.0542,
+      "step": 12569
+    },
+    {
+      "epoch": 115.32,
+      "learning_rate": 1.9686544342507644e-05,
+      "loss": 0.0572,
+      "step": 12570
+    },
+    {
+      "epoch": 115.33,
+      "learning_rate": 1.9648318042813455e-05,
+      "loss": 0.056,
+      "step": 12571
+    },
+    {
+      "epoch": 115.34,
+      "learning_rate": 1.9610091743119266e-05,
+      "loss": 0.0851,
+      "step": 12572
+    },
+    {
+      "epoch": 115.35,
+      "learning_rate": 1.9571865443425074e-05,
+      "loss": 0.0635,
+      "step": 12573
+    },
+    {
+      "epoch": 115.36,
+      "learning_rate": 1.953363914373089e-05,
+      "loss": 0.0697,
+      "step": 12574
+    },
+    {
+      "epoch": 115.37,
+      "learning_rate": 1.94954128440367e-05,
+      "loss": 0.074,
+      "step": 12575
+    },
+    {
+      "epoch": 115.38,
+      "learning_rate": 1.9457186544342507e-05,
+      "loss": 0.089,
+      "step": 12576
+    },
+    {
+      "epoch": 115.39,
+      "learning_rate": 1.9418960244648318e-05,
+      "loss": 0.0751,
+      "step": 12577
+    },
+    {
+      "epoch": 115.39,
+      "learning_rate": 1.938073394495413e-05,
+      "loss": 0.0815,
+      "step": 12578
+    },
+    {
+      "epoch": 115.4,
+      "learning_rate": 1.9342507645259937e-05,
+      "loss": 0.0921,
+      "step": 12579
+    },
+    {
+      "epoch": 115.41,
+      "learning_rate": 1.9304281345565748e-05,
+      "loss": 0.0616,
+      "step": 12580
+    },
+    {
+      "epoch": 115.42,
+      "learning_rate": 1.9266055045871563e-05,
+      "loss": 0.0563,
+      "step": 12581
+    },
+    {
+      "epoch": 115.43,
+      "learning_rate": 1.922782874617737e-05,
+      "loss": 0.0775,
+      "step": 12582
+    },
+    {
+      "epoch": 115.44,
+      "learning_rate": 1.918960244648318e-05,
+      "loss": 0.0689,
+      "step": 12583
+    },
+    {
+      "epoch": 115.45,
+      "learning_rate": 1.9151376146788992e-05,
+      "loss": 0.0449,
+      "step": 12584
+    },
+    {
+      "epoch": 115.46,
+      "learning_rate": 1.91131498470948e-05,
+      "loss": 0.0656,
+      "step": 12585
+    },
+    {
+      "epoch": 115.47,
+      "learning_rate": 1.907492354740061e-05,
+      "loss": 0.0447,
+      "step": 12586
+    },
+    {
+      "epoch": 115.48,
+      "learning_rate": 1.9036697247706422e-05,
+      "loss": 0.0708,
+      "step": 12587
+    },
+    {
+      "epoch": 115.49,
+      "learning_rate": 1.8998470948012233e-05,
+      "loss": 0.0867,
+      "step": 12588
+    },
+    {
+      "epoch": 115.5,
+      "learning_rate": 1.8960244648318044e-05,
+      "loss": 0.026,
+      "step": 12589
+    },
+    {
+      "epoch": 115.5,
+      "learning_rate": 1.8922018348623855e-05,
+      "loss": 0.1489,
+      "step": 12590
+    },
+    {
+      "epoch": 115.51,
+      "learning_rate": 1.8883792048929663e-05,
+      "loss": 0.0868,
+      "step": 12591
+    },
+    {
+      "epoch": 115.52,
+      "learning_rate": 1.8845565749235474e-05,
+      "loss": 0.0904,
+      "step": 12592
+    },
+    {
+      "epoch": 115.53,
+      "learning_rate": 1.8807339449541285e-05,
+      "loss": 0.0689,
+      "step": 12593
+    },
+    {
+      "epoch": 115.54,
+      "learning_rate": 1.8769113149847093e-05,
+      "loss": 0.109,
+      "step": 12594
+    },
+    {
+      "epoch": 115.55,
+      "learning_rate": 1.8730886850152904e-05,
+      "loss": 0.1216,
+      "step": 12595
+    },
+    {
+      "epoch": 115.56,
+      "learning_rate": 1.869266055045872e-05,
+      "loss": 0.0716,
+      "step": 12596
+    },
+    {
+      "epoch": 115.57,
+      "learning_rate": 1.8654434250764526e-05,
+      "loss": 0.1438,
+      "step": 12597
+    },
+    {
+      "epoch": 115.58,
+      "learning_rate": 1.8616207951070337e-05,
+      "loss": 0.0997,
+      "step": 12598
+    },
+    {
+      "epoch": 115.59,
+      "learning_rate": 1.8577981651376148e-05,
+      "loss": 0.1491,
+      "step": 12599
+    },
+    {
+      "epoch": 115.6,
+      "learning_rate": 1.8539755351681956e-05,
+      "loss": 0.089,
+      "step": 12600
+    },
+    {
+      "epoch": 115.61,
+      "learning_rate": 1.8501529051987767e-05,
+      "loss": 0.0648,
+      "step": 12601
+    },
+    {
+      "epoch": 115.61,
+      "learning_rate": 1.8463302752293578e-05,
+      "loss": 0.0739,
+      "step": 12602
+    },
+    {
+      "epoch": 115.62,
+      "learning_rate": 1.842507645259939e-05,
+      "loss": 0.0934,
+      "step": 12603
+    },
+    {
+      "epoch": 115.63,
+      "learning_rate": 1.83868501529052e-05,
+      "loss": 0.0991,
+      "step": 12604
+    },
+    {
+      "epoch": 115.64,
+      "learning_rate": 1.834862385321101e-05,
+      "loss": 0.1118,
+      "step": 12605
+    },
+    {
+      "epoch": 115.65,
+      "learning_rate": 1.831039755351682e-05,
+      "loss": 0.0655,
+      "step": 12606
+    },
+    {
+      "epoch": 115.66,
+      "learning_rate": 1.827217125382263e-05,
+      "loss": 0.1537,
+      "step": 12607
+    },
+    {
+      "epoch": 115.67,
+      "learning_rate": 1.823394495412844e-05,
+      "loss": 0.0897,
+      "step": 12608
+    },
+    {
+      "epoch": 115.68,
+      "learning_rate": 1.819571865443425e-05,
+      "loss": 0.0516,
+      "step": 12609
+    },
+    {
+      "epoch": 115.69,
+      "learning_rate": 1.8157492354740063e-05,
+      "loss": 0.0767,
+      "step": 12610
+    },
+    {
+      "epoch": 115.7,
+      "learning_rate": 1.8119266055045874e-05,
+      "loss": 0.0908,
+      "step": 12611
+    },
+    {
+      "epoch": 115.71,
+      "learning_rate": 1.8081039755351682e-05,
+      "loss": 0.0929,
+      "step": 12612
+    },
+    {
+      "epoch": 115.72,
+      "learning_rate": 1.8042813455657493e-05,
+      "loss": 0.0285,
+      "step": 12613
+    },
+    {
+      "epoch": 115.72,
+      "learning_rate": 1.8004587155963304e-05,
+      "loss": 0.0591,
+      "step": 12614
+    },
+    {
+      "epoch": 115.73,
+      "learning_rate": 1.796636085626911e-05,
+      "loss": 0.0719,
+      "step": 12615
+    },
+    {
+      "epoch": 115.74,
+      "learning_rate": 1.7928134556574923e-05,
+      "loss": 0.0289,
+      "step": 12616
+    },
+    {
+      "epoch": 115.75,
+      "learning_rate": 1.7889908256880734e-05,
+      "loss": 0.1091,
+      "step": 12617
+    },
+    {
+      "epoch": 115.76,
+      "learning_rate": 1.7851681957186545e-05,
+      "loss": 0.1162,
+      "step": 12618
+    },
+    {
+      "epoch": 115.77,
+      "learning_rate": 1.7813455657492356e-05,
+      "loss": 0.1099,
+      "step": 12619
+    },
+    {
+      "epoch": 115.78,
+      "learning_rate": 1.7775229357798167e-05,
+      "loss": 0.0478,
+      "step": 12620
+    },
+    {
+      "epoch": 115.79,
+      "learning_rate": 1.7737003058103975e-05,
+      "loss": 0.1074,
+      "step": 12621
+    },
+    {
+      "epoch": 115.8,
+      "learning_rate": 1.7698776758409786e-05,
+      "loss": 0.0839,
+      "step": 12622
+    },
+    {
+      "epoch": 115.81,
+      "learning_rate": 1.7660550458715597e-05,
+      "loss": 0.0604,
+      "step": 12623
+    },
+    {
+      "epoch": 115.82,
+      "learning_rate": 1.7622324159021404e-05,
+      "loss": 0.0761,
+      "step": 12624
+    },
+    {
+      "epoch": 115.83,
+      "learning_rate": 1.758409785932722e-05,
+      "loss": 0.0541,
+      "step": 12625
+    },
+    {
+      "epoch": 115.83,
+      "learning_rate": 1.754587155963303e-05,
+      "loss": 0.1221,
+      "step": 12626
+    },
+    {
+      "epoch": 115.84,
+      "learning_rate": 1.7507645259938838e-05,
+      "loss": 0.099,
+      "step": 12627
+    },
+    {
+      "epoch": 115.85,
+      "learning_rate": 1.746941896024465e-05,
+      "loss": 0.0431,
+      "step": 12628
+    },
+    {
+      "epoch": 115.86,
+      "learning_rate": 1.743119266055046e-05,
+      "loss": 0.0609,
+      "step": 12629
+    },
+    {
+      "epoch": 115.87,
+      "learning_rate": 1.7392966360856267e-05,
+      "loss": 0.0464,
+      "step": 12630
+    },
+    {
+      "epoch": 115.88,
+      "learning_rate": 1.735474006116208e-05,
+      "loss": 0.1192,
+      "step": 12631
+    },
+    {
+      "epoch": 115.89,
+      "learning_rate": 1.731651376146789e-05,
+      "loss": 0.0818,
+      "step": 12632
+    },
+    {
+      "epoch": 115.9,
+      "learning_rate": 1.72782874617737e-05,
+      "loss": 0.0886,
+      "step": 12633
+    },
+    {
+      "epoch": 115.91,
+      "learning_rate": 1.724006116207951e-05,
+      "loss": 0.099,
+      "step": 12634
+    },
+    {
+      "epoch": 115.92,
+      "learning_rate": 1.7201834862385323e-05,
+      "loss": 0.0556,
+      "step": 12635
+    },
+    {
+      "epoch": 115.93,
+      "learning_rate": 1.716360856269113e-05,
+      "loss": 0.0476,
+      "step": 12636
+    },
+    {
+      "epoch": 115.94,
+      "learning_rate": 1.712538226299694e-05,
+      "loss": 0.0718,
+      "step": 12637
+    },
+    {
+      "epoch": 115.94,
+      "learning_rate": 1.7087155963302753e-05,
+      "loss": 0.1165,
+      "step": 12638
+    },
+    {
+      "epoch": 115.95,
+      "learning_rate": 1.704892966360856e-05,
+      "loss": 0.074,
+      "step": 12639
+    },
+    {
+      "epoch": 115.96,
+      "learning_rate": 1.7010703363914375e-05,
+      "loss": 0.1273,
+      "step": 12640
+    },
+    {
+      "epoch": 115.97,
+      "learning_rate": 1.6972477064220186e-05,
+      "loss": 0.1123,
+      "step": 12641
+    },
+    {
+      "epoch": 115.98,
+      "learning_rate": 1.6934250764525993e-05,
+      "loss": 0.1016,
+      "step": 12642
+    },
+    {
+      "epoch": 115.99,
+      "learning_rate": 1.6896024464831804e-05,
+      "loss": 0.0672,
+      "step": 12643
+    },
+    {
+      "epoch": 116.0,
+      "learning_rate": 1.6857798165137616e-05,
+      "loss": 0.0922,
+      "step": 12644
+    },
+    {
+      "epoch": 116.01,
+      "learning_rate": 1.6819571865443423e-05,
+      "loss": 0.136,
+      "step": 12645
+    },
+    {
+      "epoch": 116.02,
+      "learning_rate": 1.6781345565749234e-05,
+      "loss": 0.1233,
+      "step": 12646
+    },
+    {
+      "epoch": 116.03,
+      "learning_rate": 1.674311926605505e-05,
+      "loss": 0.0797,
+      "step": 12647
+    },
+    {
+      "epoch": 116.04,
+      "learning_rate": 1.6704892966360856e-05,
+      "loss": 0.1553,
+      "step": 12648
+    },
+    {
+      "epoch": 116.05,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 0.0748,
+      "step": 12649
+    },
+    {
+      "epoch": 116.06,
+      "learning_rate": 1.662844036697248e-05,
+      "loss": 0.0715,
+      "step": 12650
+    },
+    {
+      "epoch": 116.06,
+      "learning_rate": 1.6590214067278286e-05,
+      "loss": 0.0724,
+      "step": 12651
+    },
+    {
+      "epoch": 116.07,
+      "learning_rate": 1.6551987767584097e-05,
+      "loss": 0.0711,
+      "step": 12652
+    },
+    {
+      "epoch": 116.08,
+      "learning_rate": 1.651376146788991e-05,
+      "loss": 0.0787,
+      "step": 12653
+    },
+    {
+      "epoch": 116.09,
+      "learning_rate": 1.6475535168195716e-05,
+      "loss": 0.0587,
+      "step": 12654
+    },
+    {
+      "epoch": 116.1,
+      "learning_rate": 1.643730886850153e-05,
+      "loss": 0.0704,
+      "step": 12655
+    },
+    {
+      "epoch": 116.11,
+      "learning_rate": 1.639908256880734e-05,
+      "loss": 0.0534,
+      "step": 12656
+    },
+    {
+      "epoch": 116.12,
+      "learning_rate": 1.636085626911315e-05,
+      "loss": 0.1022,
+      "step": 12657
+    },
+    {
+      "epoch": 116.13,
+      "learning_rate": 1.632262996941896e-05,
+      "loss": 0.1064,
+      "step": 12658
+    },
+    {
+      "epoch": 116.14,
+      "learning_rate": 1.628440366972477e-05,
+      "loss": 0.0746,
+      "step": 12659
+    },
+    {
+      "epoch": 116.15,
+      "learning_rate": 1.624617737003058e-05,
+      "loss": 0.0802,
+      "step": 12660
+    },
+    {
+      "epoch": 116.16,
+      "learning_rate": 1.620795107033639e-05,
+      "loss": 0.0972,
+      "step": 12661
+    },
+    {
+      "epoch": 116.17,
+      "learning_rate": 1.6169724770642205e-05,
+      "loss": 0.0671,
+      "step": 12662
+    },
+    {
+      "epoch": 116.17,
+      "learning_rate": 1.6131498470948012e-05,
+      "loss": 0.0851,
+      "step": 12663
+    },
+    {
+      "epoch": 116.18,
+      "learning_rate": 1.6093272171253823e-05,
+      "loss": 0.0964,
+      "step": 12664
+    },
+    {
+      "epoch": 116.19,
+      "learning_rate": 1.6055045871559634e-05,
+      "loss": 0.0691,
+      "step": 12665
+    },
+    {
+      "epoch": 116.2,
+      "learning_rate": 1.6016819571865442e-05,
+      "loss": 0.1287,
+      "step": 12666
+    },
+    {
+      "epoch": 116.21,
+      "learning_rate": 1.5978593272171253e-05,
+      "loss": 0.0968,
+      "step": 12667
+    },
+    {
+      "epoch": 116.22,
+      "learning_rate": 1.5940366972477064e-05,
+      "loss": 0.1021,
+      "step": 12668
+    },
+    {
+      "epoch": 116.23,
+      "learning_rate": 1.5902140672782875e-05,
+      "loss": 0.0877,
+      "step": 12669
+    },
+    {
+      "epoch": 116.24,
+      "learning_rate": 1.5863914373088686e-05,
+      "loss": 0.0385,
+      "step": 12670
+    },
+    {
+      "epoch": 116.25,
+      "learning_rate": 1.5825688073394497e-05,
+      "loss": 0.0425,
+      "step": 12671
+    },
+    {
+      "epoch": 116.26,
+      "learning_rate": 1.5787461773700305e-05,
+      "loss": 0.1641,
+      "step": 12672
+    },
+    {
+      "epoch": 116.27,
+      "learning_rate": 1.5749235474006116e-05,
+      "loss": 0.115,
+      "step": 12673
+    },
+    {
+      "epoch": 116.28,
+      "learning_rate": 1.5711009174311927e-05,
+      "loss": 0.0985,
+      "step": 12674
+    },
+    {
+      "epoch": 116.28,
+      "learning_rate": 1.5672782874617735e-05,
+      "loss": 0.1278,
+      "step": 12675
+    },
+    {
+      "epoch": 116.29,
+      "learning_rate": 1.5634556574923546e-05,
+      "loss": 0.0604,
+      "step": 12676
+    },
+    {
+      "epoch": 116.3,
+      "learning_rate": 1.559633027522936e-05,
+      "loss": 0.1152,
+      "step": 12677
+    },
+    {
+      "epoch": 116.31,
+      "learning_rate": 1.5558103975535168e-05,
+      "loss": 0.1321,
+      "step": 12678
+    },
+    {
+      "epoch": 116.32,
+      "learning_rate": 1.551987767584098e-05,
+      "loss": 0.0907,
+      "step": 12679
+    },
+    {
+      "epoch": 116.33,
+      "learning_rate": 1.548165137614679e-05,
+      "loss": 0.0623,
+      "step": 12680
+    },
+    {
+      "epoch": 116.34,
+      "learning_rate": 1.54434250764526e-05,
+      "loss": 0.0786,
+      "step": 12681
+    },
+    {
+      "epoch": 116.35,
+      "learning_rate": 1.540519877675841e-05,
+      "loss": 0.0644,
+      "step": 12682
+    },
+    {
+      "epoch": 116.36,
+      "learning_rate": 1.536697247706422e-05,
+      "loss": 0.0655,
+      "step": 12683
+    },
+    {
+      "epoch": 116.37,
+      "learning_rate": 1.532874617737003e-05,
+      "loss": 0.0964,
+      "step": 12684
+    },
+    {
+      "epoch": 116.38,
+      "learning_rate": 1.5290519877675842e-05,
+      "loss": 0.0788,
+      "step": 12685
+    },
+    {
+      "epoch": 116.39,
+      "learning_rate": 1.5252293577981653e-05,
+      "loss": 0.13,
+      "step": 12686
+    },
+    {
+      "epoch": 116.39,
+      "learning_rate": 1.5214067278287462e-05,
+      "loss": 0.083,
+      "step": 12687
+    },
+    {
+      "epoch": 116.4,
+      "learning_rate": 1.5175840978593272e-05,
+      "loss": 0.0699,
+      "step": 12688
+    },
+    {
+      "epoch": 116.41,
+      "learning_rate": 1.5137614678899083e-05,
+      "loss": 0.1139,
+      "step": 12689
+    },
+    {
+      "epoch": 116.42,
+      "learning_rate": 1.5099388379204894e-05,
+      "loss": 0.0565,
+      "step": 12690
+    },
+    {
+      "epoch": 116.43,
+      "learning_rate": 1.5061162079510703e-05,
+      "loss": 0.0785,
+      "step": 12691
+    },
+    {
+      "epoch": 116.44,
+      "learning_rate": 1.5022935779816514e-05,
+      "loss": 0.0577,
+      "step": 12692
+    },
+    {
+      "epoch": 116.45,
+      "learning_rate": 1.4984709480122325e-05,
+      "loss": 0.0782,
+      "step": 12693
+    },
+    {
+      "epoch": 116.46,
+      "learning_rate": 1.4946483180428135e-05,
+      "loss": 0.1213,
+      "step": 12694
+    },
+    {
+      "epoch": 116.47,
+      "learning_rate": 1.4908256880733946e-05,
+      "loss": 0.0608,
+      "step": 12695
+    },
+    {
+      "epoch": 116.48,
+      "learning_rate": 1.4870030581039755e-05,
+      "loss": 0.0515,
+      "step": 12696
+    },
+    {
+      "epoch": 116.49,
+      "learning_rate": 1.4831804281345566e-05,
+      "loss": 0.0571,
+      "step": 12697
+    },
+    {
+      "epoch": 116.5,
+      "learning_rate": 1.4793577981651377e-05,
+      "loss": 0.0421,
+      "step": 12698
+    },
+    {
+      "epoch": 116.5,
+      "learning_rate": 1.4755351681957187e-05,
+      "loss": 0.109,
+      "step": 12699
+    },
+    {
+      "epoch": 116.51,
+      "learning_rate": 1.4717125382262996e-05,
+      "loss": 0.138,
+      "step": 12700
+    },
+    {
+      "epoch": 116.52,
+      "learning_rate": 1.4678899082568809e-05,
+      "loss": 0.0833,
+      "step": 12701
+    },
+    {
+      "epoch": 116.53,
+      "learning_rate": 1.4640672782874618e-05,
+      "loss": 0.0685,
+      "step": 12702
+    },
+    {
+      "epoch": 116.54,
+      "learning_rate": 1.4602446483180428e-05,
+      "loss": 0.1376,
+      "step": 12703
+    },
+    {
+      "epoch": 116.55,
+      "learning_rate": 1.456422018348624e-05,
+      "loss": 0.1004,
+      "step": 12704
+    },
+    {
+      "epoch": 116.56,
+      "learning_rate": 1.452599388379205e-05,
+      "loss": 0.0915,
+      "step": 12705
+    },
+    {
+      "epoch": 116.57,
+      "learning_rate": 1.4487767584097859e-05,
+      "loss": 0.0668,
+      "step": 12706
+    },
+    {
+      "epoch": 116.58,
+      "learning_rate": 1.444954128440367e-05,
+      "loss": 0.1055,
+      "step": 12707
+    },
+    {
+      "epoch": 116.59,
+      "learning_rate": 1.4411314984709481e-05,
+      "loss": 0.0677,
+      "step": 12708
+    },
+    {
+      "epoch": 116.6,
+      "learning_rate": 1.437308868501529e-05,
+      "loss": 0.0992,
+      "step": 12709
+    },
+    {
+      "epoch": 116.61,
+      "learning_rate": 1.4334862385321102e-05,
+      "loss": 0.0618,
+      "step": 12710
+    },
+    {
+      "epoch": 116.61,
+      "learning_rate": 1.4296636085626911e-05,
+      "loss": 0.0965,
+      "step": 12711
+    },
+    {
+      "epoch": 116.62,
+      "learning_rate": 1.4258409785932722e-05,
+      "loss": 0.0212,
+      "step": 12712
+    },
+    {
+      "epoch": 116.63,
+      "learning_rate": 1.4220183486238533e-05,
+      "loss": 0.1372,
+      "step": 12713
+    },
+    {
+      "epoch": 116.64,
+      "learning_rate": 1.4181957186544343e-05,
+      "loss": 0.0365,
+      "step": 12714
+    },
+    {
+      "epoch": 116.65,
+      "learning_rate": 1.4143730886850154e-05,
+      "loss": 0.0879,
+      "step": 12715
+    },
+    {
+      "epoch": 116.66,
+      "learning_rate": 1.4105504587155965e-05,
+      "loss": 0.096,
+      "step": 12716
+    },
+    {
+      "epoch": 116.67,
+      "learning_rate": 1.4067278287461774e-05,
+      "loss": 0.1085,
+      "step": 12717
+    },
+    {
+      "epoch": 116.68,
+      "learning_rate": 1.4029051987767583e-05,
+      "loss": 0.0803,
+      "step": 12718
+    },
+    {
+      "epoch": 116.69,
+      "learning_rate": 1.3990825688073396e-05,
+      "loss": 0.0967,
+      "step": 12719
+    },
+    {
+      "epoch": 116.7,
+      "learning_rate": 1.3952599388379206e-05,
+      "loss": 0.0919,
+      "step": 12720
+    },
+    {
+      "epoch": 116.71,
+      "learning_rate": 1.3914373088685015e-05,
+      "loss": 0.0623,
+      "step": 12721
+    },
+    {
+      "epoch": 116.72,
+      "learning_rate": 1.3876146788990826e-05,
+      "loss": 0.0482,
+      "step": 12722
+    },
+    {
+      "epoch": 116.72,
+      "learning_rate": 1.3837920489296637e-05,
+      "loss": 0.0524,
+      "step": 12723
+    },
+    {
+      "epoch": 116.73,
+      "learning_rate": 1.3799694189602446e-05,
+      "loss": 0.1076,
+      "step": 12724
+    },
+    {
+      "epoch": 116.74,
+      "learning_rate": 1.3761467889908258e-05,
+      "loss": 0.0682,
+      "step": 12725
+    },
+    {
+      "epoch": 116.75,
+      "learning_rate": 1.3723241590214069e-05,
+      "loss": 0.1372,
+      "step": 12726
+    },
+    {
+      "epoch": 116.76,
+      "learning_rate": 1.3685015290519878e-05,
+      "loss": 0.1018,
+      "step": 12727
+    },
+    {
+      "epoch": 116.77,
+      "learning_rate": 1.3646788990825689e-05,
+      "loss": 0.1399,
+      "step": 12728
+    },
+    {
+      "epoch": 116.78,
+      "learning_rate": 1.3608562691131498e-05,
+      "loss": 0.0652,
+      "step": 12729
+    },
+    {
+      "epoch": 116.79,
+      "learning_rate": 1.357033639143731e-05,
+      "loss": 0.0827,
+      "step": 12730
+    },
+    {
+      "epoch": 116.8,
+      "learning_rate": 1.353211009174312e-05,
+      "loss": 0.0919,
+      "step": 12731
+    },
+    {
+      "epoch": 116.81,
+      "learning_rate": 1.349388379204893e-05,
+      "loss": 0.0833,
+      "step": 12732
+    },
+    {
+      "epoch": 116.82,
+      "learning_rate": 1.345565749235474e-05,
+      "loss": 0.0818,
+      "step": 12733
+    },
+    {
+      "epoch": 116.83,
+      "learning_rate": 1.3417431192660552e-05,
+      "loss": 0.1084,
+      "step": 12734
+    },
+    {
+      "epoch": 116.83,
+      "learning_rate": 1.3379204892966361e-05,
+      "loss": 0.0712,
+      "step": 12735
+    },
+    {
+      "epoch": 116.84,
+      "learning_rate": 1.334097859327217e-05,
+      "loss": 0.0977,
+      "step": 12736
+    },
+    {
+      "epoch": 116.85,
+      "learning_rate": 1.3302752293577984e-05,
+      "loss": 0.0766,
+      "step": 12737
+    },
+    {
+      "epoch": 116.86,
+      "learning_rate": 1.3264525993883793e-05,
+      "loss": 0.0824,
+      "step": 12738
+    },
+    {
+      "epoch": 116.87,
+      "learning_rate": 1.3226299694189602e-05,
+      "loss": 0.0724,
+      "step": 12739
+    },
+    {
+      "epoch": 116.88,
+      "learning_rate": 1.3188073394495413e-05,
+      "loss": 0.0758,
+      "step": 12740
+    },
+    {
+      "epoch": 116.89,
+      "learning_rate": 1.3149847094801224e-05,
+      "loss": 0.0559,
+      "step": 12741
+    },
+    {
+      "epoch": 116.9,
+      "learning_rate": 1.3111620795107034e-05,
+      "loss": 0.038,
+      "step": 12742
+    },
+    {
+      "epoch": 116.91,
+      "learning_rate": 1.3073394495412845e-05,
+      "loss": 0.0525,
+      "step": 12743
+    },
+    {
+      "epoch": 116.92,
+      "learning_rate": 1.3035168195718654e-05,
+      "loss": 0.1185,
+      "step": 12744
+    },
+    {
+      "epoch": 116.93,
+      "learning_rate": 1.2996941896024465e-05,
+      "loss": 0.1011,
+      "step": 12745
+    },
+    {
+      "epoch": 116.94,
+      "learning_rate": 1.2958715596330276e-05,
+      "loss": 0.045,
+      "step": 12746
+    },
+    {
+      "epoch": 116.94,
+      "learning_rate": 1.2920489296636086e-05,
+      "loss": 0.0686,
+      "step": 12747
+    },
+    {
+      "epoch": 116.95,
+      "learning_rate": 1.2882262996941895e-05,
+      "loss": 0.0831,
+      "step": 12748
+    },
+    {
+      "epoch": 116.96,
+      "learning_rate": 1.2844036697247708e-05,
+      "loss": 0.0999,
+      "step": 12749
+    },
+    {
+      "epoch": 116.97,
+      "learning_rate": 1.2805810397553517e-05,
+      "loss": 0.0561,
+      "step": 12750
+    },
+    {
+      "epoch": 116.98,
+      "learning_rate": 1.2767584097859327e-05,
+      "loss": 0.0541,
+      "step": 12751
+    },
+    {
+      "epoch": 116.99,
+      "learning_rate": 1.272935779816514e-05,
+      "loss": 0.0299,
+      "step": 12752
+    },
+    {
+      "epoch": 117.0,
+      "learning_rate": 1.2691131498470949e-05,
+      "loss": 0.0805,
+      "step": 12753
+    },
+    {
+      "epoch": 117.01,
+      "learning_rate": 1.2652905198776758e-05,
+      "loss": 0.1076,
+      "step": 12754
+    },
+    {
+      "epoch": 117.02,
+      "learning_rate": 1.2614678899082569e-05,
+      "loss": 0.0879,
+      "step": 12755
+    },
+    {
+      "epoch": 117.03,
+      "learning_rate": 1.257645259938838e-05,
+      "loss": 0.0783,
+      "step": 12756
+    },
+    {
+      "epoch": 117.04,
+      "learning_rate": 1.253822629969419e-05,
+      "loss": 0.0625,
+      "step": 12757
+    },
+    {
+      "epoch": 117.05,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0858,
+      "step": 12758
+    },
+    {
+      "epoch": 117.06,
+      "learning_rate": 1.246177370030581e-05,
+      "loss": 0.1333,
+      "step": 12759
+    },
+    {
+      "epoch": 117.06,
+      "learning_rate": 1.2423547400611621e-05,
+      "loss": 0.0948,
+      "step": 12760
+    },
+    {
+      "epoch": 117.07,
+      "learning_rate": 1.2385321100917432e-05,
+      "loss": 0.0759,
+      "step": 12761
+    },
+    {
+      "epoch": 117.08,
+      "learning_rate": 1.2347094801223241e-05,
+      "loss": 0.0978,
+      "step": 12762
+    },
+    {
+      "epoch": 117.09,
+      "learning_rate": 1.2308868501529053e-05,
+      "loss": 0.1331,
+      "step": 12763
+    },
+    {
+      "epoch": 117.1,
+      "learning_rate": 1.2270642201834864e-05,
+      "loss": 0.0754,
+      "step": 12764
+    },
+    {
+      "epoch": 117.11,
+      "learning_rate": 1.2232415902140673e-05,
+      "loss": 0.0706,
+      "step": 12765
+    },
+    {
+      "epoch": 117.12,
+      "learning_rate": 1.2194189602446482e-05,
+      "loss": 0.1087,
+      "step": 12766
+    },
+    {
+      "epoch": 117.13,
+      "learning_rate": 1.2155963302752295e-05,
+      "loss": 0.0646,
+      "step": 12767
+    },
+    {
+      "epoch": 117.14,
+      "learning_rate": 1.2117737003058104e-05,
+      "loss": 0.0838,
+      "step": 12768
+    },
+    {
+      "epoch": 117.15,
+      "learning_rate": 1.2079510703363914e-05,
+      "loss": 0.1438,
+      "step": 12769
+    },
+    {
+      "epoch": 117.16,
+      "learning_rate": 1.2041284403669725e-05,
+      "loss": 0.0651,
+      "step": 12770
+    },
+    {
+      "epoch": 117.17,
+      "learning_rate": 1.2003058103975536e-05,
+      "loss": 0.1033,
+      "step": 12771
+    },
+    {
+      "epoch": 117.17,
+      "learning_rate": 1.1964831804281345e-05,
+      "loss": 0.1231,
+      "step": 12772
+    },
+    {
+      "epoch": 117.18,
+      "learning_rate": 1.1926605504587156e-05,
+      "loss": 0.09,
+      "step": 12773
+    },
+    {
+      "epoch": 117.19,
+      "learning_rate": 1.1888379204892967e-05,
+      "loss": 0.0588,
+      "step": 12774
+    },
+    {
+      "epoch": 117.2,
+      "learning_rate": 1.1850152905198777e-05,
+      "loss": 0.1278,
+      "step": 12775
+    },
+    {
+      "epoch": 117.21,
+      "learning_rate": 1.1811926605504588e-05,
+      "loss": 0.0324,
+      "step": 12776
+    },
+    {
+      "epoch": 117.22,
+      "learning_rate": 1.1773700305810397e-05,
+      "loss": 0.0735,
+      "step": 12777
+    },
+    {
+      "epoch": 117.23,
+      "learning_rate": 1.1735474006116208e-05,
+      "loss": 0.13,
+      "step": 12778
+    },
+    {
+      "epoch": 117.24,
+      "learning_rate": 1.169724770642202e-05,
+      "loss": 0.0992,
+      "step": 12779
+    },
+    {
+      "epoch": 117.25,
+      "learning_rate": 1.1659021406727829e-05,
+      "loss": 0.0153,
+      "step": 12780
+    },
+    {
+      "epoch": 117.26,
+      "learning_rate": 1.1620795107033638e-05,
+      "loss": 0.1697,
+      "step": 12781
+    },
+    {
+      "epoch": 117.27,
+      "learning_rate": 1.1582568807339451e-05,
+      "loss": 0.0935,
+      "step": 12782
+    },
+    {
+      "epoch": 117.28,
+      "learning_rate": 1.154434250764526e-05,
+      "loss": 0.0788,
+      "step": 12783
+    },
+    {
+      "epoch": 117.28,
+      "learning_rate": 1.150611620795107e-05,
+      "loss": 0.1064,
+      "step": 12784
+    },
+    {
+      "epoch": 117.29,
+      "learning_rate": 1.1467889908256882e-05,
+      "loss": 0.1112,
+      "step": 12785
+    },
+    {
+      "epoch": 117.3,
+      "learning_rate": 1.1429663608562692e-05,
+      "loss": 0.1042,
+      "step": 12786
+    },
+    {
+      "epoch": 117.31,
+      "learning_rate": 1.1391437308868501e-05,
+      "loss": 0.0636,
+      "step": 12787
+    },
+    {
+      "epoch": 117.32,
+      "learning_rate": 1.1353211009174312e-05,
+      "loss": 0.0847,
+      "step": 12788
+    },
+    {
+      "epoch": 117.33,
+      "learning_rate": 1.1314984709480123e-05,
+      "loss": 0.1119,
+      "step": 12789
+    },
+    {
+      "epoch": 117.34,
+      "learning_rate": 1.1276758409785933e-05,
+      "loss": 0.0956,
+      "step": 12790
+    },
+    {
+      "epoch": 117.35,
+      "learning_rate": 1.1238532110091744e-05,
+      "loss": 0.0796,
+      "step": 12791
+    },
+    {
+      "epoch": 117.36,
+      "learning_rate": 1.1200305810397553e-05,
+      "loss": 0.0699,
+      "step": 12792
+    },
+    {
+      "epoch": 117.37,
+      "learning_rate": 1.1162079510703364e-05,
+      "loss": 0.0687,
+      "step": 12793
+    },
+    {
+      "epoch": 117.38,
+      "learning_rate": 1.1123853211009175e-05,
+      "loss": 0.0704,
+      "step": 12794
+    },
+    {
+      "epoch": 117.39,
+      "learning_rate": 1.1085626911314985e-05,
+      "loss": 0.0601,
+      "step": 12795
+    },
+    {
+      "epoch": 117.39,
+      "learning_rate": 1.1047400611620796e-05,
+      "loss": 0.0587,
+      "step": 12796
+    },
+    {
+      "epoch": 117.4,
+      "learning_rate": 1.1009174311926607e-05,
+      "loss": 0.0502,
+      "step": 12797
+    },
+    {
+      "epoch": 117.41,
+      "learning_rate": 1.0970948012232416e-05,
+      "loss": 0.0955,
+      "step": 12798
+    },
+    {
+      "epoch": 117.42,
+      "learning_rate": 1.0932721712538225e-05,
+      "loss": 0.1074,
+      "step": 12799
+    },
+    {
+      "epoch": 117.43,
+      "learning_rate": 1.0894495412844038e-05,
+      "loss": 0.0912,
+      "step": 12800
+    },
+    {
+      "epoch": 117.44,
+      "learning_rate": 1.0856269113149848e-05,
+      "loss": 0.0716,
+      "step": 12801
+    },
+    {
+      "epoch": 117.45,
+      "learning_rate": 1.0818042813455657e-05,
+      "loss": 0.0292,
+      "step": 12802
+    },
+    {
+      "epoch": 117.46,
+      "learning_rate": 1.0779816513761468e-05,
+      "loss": 0.0392,
+      "step": 12803
+    },
+    {
+      "epoch": 117.47,
+      "learning_rate": 1.0741590214067279e-05,
+      "loss": 0.0661,
+      "step": 12804
+    },
+    {
+      "epoch": 117.48,
+      "learning_rate": 1.0703363914373088e-05,
+      "loss": 0.0421,
+      "step": 12805
+    },
+    {
+      "epoch": 117.49,
+      "learning_rate": 1.06651376146789e-05,
+      "loss": 0.0701,
+      "step": 12806
+    },
+    {
+      "epoch": 117.5,
+      "learning_rate": 1.062691131498471e-05,
+      "loss": 0.0272,
+      "step": 12807
+    },
+    {
+      "epoch": 117.5,
+      "learning_rate": 1.058868501529052e-05,
+      "loss": 0.1286,
+      "step": 12808
+    },
+    {
+      "epoch": 117.51,
+      "learning_rate": 1.0550458715596331e-05,
+      "loss": 0.0922,
+      "step": 12809
+    },
+    {
+      "epoch": 117.52,
+      "learning_rate": 1.051223241590214e-05,
+      "loss": 0.1187,
+      "step": 12810
+    },
+    {
+      "epoch": 117.53,
+      "learning_rate": 1.0474006116207951e-05,
+      "loss": 0.1281,
+      "step": 12811
+    },
+    {
+      "epoch": 117.54,
+      "learning_rate": 1.0435779816513762e-05,
+      "loss": 0.1594,
+      "step": 12812
+    },
+    {
+      "epoch": 117.55,
+      "learning_rate": 1.0397553516819572e-05,
+      "loss": 0.093,
+      "step": 12813
+    },
+    {
+      "epoch": 117.56,
+      "learning_rate": 1.0359327217125381e-05,
+      "loss": 0.1401,
+      "step": 12814
+    },
+    {
+      "epoch": 117.57,
+      "learning_rate": 1.0321100917431194e-05,
+      "loss": 0.074,
+      "step": 12815
+    },
+    {
+      "epoch": 117.58,
+      "learning_rate": 1.0282874617737003e-05,
+      "loss": 0.0885,
+      "step": 12816
+    },
+    {
+      "epoch": 117.59,
+      "learning_rate": 1.0244648318042813e-05,
+      "loss": 0.0818,
+      "step": 12817
+    },
+    {
+      "epoch": 117.6,
+      "learning_rate": 1.0206422018348625e-05,
+      "loss": 0.0798,
+      "step": 12818
+    },
+    {
+      "epoch": 117.61,
+      "learning_rate": 1.0168195718654435e-05,
+      "loss": 0.0377,
+      "step": 12819
+    },
+    {
+      "epoch": 117.61,
+      "learning_rate": 1.0129969418960244e-05,
+      "loss": 0.0575,
+      "step": 12820
+    },
+    {
+      "epoch": 117.62,
+      "learning_rate": 1.0091743119266055e-05,
+      "loss": 0.0954,
+      "step": 12821
+    },
+    {
+      "epoch": 117.63,
+      "learning_rate": 1.0053516819571866e-05,
+      "loss": 0.0845,
+      "step": 12822
+    },
+    {
+      "epoch": 117.64,
+      "learning_rate": 1.0015290519877676e-05,
+      "loss": 0.0616,
+      "step": 12823
+    },
+    {
+      "epoch": 117.65,
+      "learning_rate": 9.977064220183487e-06,
+      "loss": 0.0917,
+      "step": 12824
+    },
+    {
+      "epoch": 117.66,
+      "learning_rate": 9.938837920489296e-06,
+      "loss": 0.0772,
+      "step": 12825
+    },
+    {
+      "epoch": 117.67,
+      "learning_rate": 9.900611620795107e-06,
+      "loss": 0.0693,
+      "step": 12826
+    },
+    {
+      "epoch": 117.68,
+      "learning_rate": 9.862385321100918e-06,
+      "loss": 0.0445,
+      "step": 12827
+    },
+    {
+      "epoch": 117.69,
+      "learning_rate": 9.824159021406728e-06,
+      "loss": 0.0548,
+      "step": 12828
+    },
+    {
+      "epoch": 117.7,
+      "learning_rate": 9.785932721712537e-06,
+      "loss": 0.097,
+      "step": 12829
+    },
+    {
+      "epoch": 117.71,
+      "learning_rate": 9.74770642201835e-06,
+      "loss": 0.0884,
+      "step": 12830
+    },
+    {
+      "epoch": 117.72,
+      "learning_rate": 9.709480122324159e-06,
+      "loss": 0.0395,
+      "step": 12831
+    },
+    {
+      "epoch": 117.72,
+      "learning_rate": 9.671253822629969e-06,
+      "loss": 0.0365,
+      "step": 12832
+    },
+    {
+      "epoch": 117.73,
+      "learning_rate": 9.633027522935781e-06,
+      "loss": 0.0208,
+      "step": 12833
+    },
+    {
+      "epoch": 117.74,
+      "learning_rate": 9.59480122324159e-06,
+      "loss": 0.0822,
+      "step": 12834
+    },
+    {
+      "epoch": 117.75,
+      "learning_rate": 9.5565749235474e-06,
+      "loss": 0.1197,
+      "step": 12835
+    },
+    {
+      "epoch": 117.76,
+      "learning_rate": 9.518348623853211e-06,
+      "loss": 0.108,
+      "step": 12836
+    },
+    {
+      "epoch": 117.77,
+      "learning_rate": 9.480122324159022e-06,
+      "loss": 0.0714,
+      "step": 12837
+    },
+    {
+      "epoch": 117.78,
+      "learning_rate": 9.441896024464832e-06,
+      "loss": 0.0936,
+      "step": 12838
+    },
+    {
+      "epoch": 117.79,
+      "learning_rate": 9.403669724770643e-06,
+      "loss": 0.0858,
+      "step": 12839
+    },
+    {
+      "epoch": 117.8,
+      "learning_rate": 9.365443425076452e-06,
+      "loss": 0.0995,
+      "step": 12840
+    },
+    {
+      "epoch": 117.81,
+      "learning_rate": 9.327217125382263e-06,
+      "loss": 0.0994,
+      "step": 12841
+    },
+    {
+      "epoch": 117.82,
+      "learning_rate": 9.288990825688074e-06,
+      "loss": 0.0461,
+      "step": 12842
+    },
+    {
+      "epoch": 117.83,
+      "learning_rate": 9.250764525993883e-06,
+      "loss": 0.0631,
+      "step": 12843
+    },
+    {
+      "epoch": 117.83,
+      "learning_rate": 9.212538226299695e-06,
+      "loss": 0.0922,
+      "step": 12844
+    },
+    {
+      "epoch": 117.84,
+      "learning_rate": 9.174311926605506e-06,
+      "loss": 0.0548,
+      "step": 12845
+    },
+    {
+      "epoch": 117.85,
+      "learning_rate": 9.136085626911315e-06,
+      "loss": 0.1024,
+      "step": 12846
+    },
+    {
+      "epoch": 117.86,
+      "learning_rate": 9.097859327217124e-06,
+      "loss": 0.0951,
+      "step": 12847
+    },
+    {
+      "epoch": 117.87,
+      "learning_rate": 9.059633027522937e-06,
+      "loss": 0.0716,
+      "step": 12848
+    },
+    {
+      "epoch": 117.88,
+      "learning_rate": 9.021406727828746e-06,
+      "loss": 0.0778,
+      "step": 12849
+    },
+    {
+      "epoch": 117.89,
+      "learning_rate": 8.983180428134556e-06,
+      "loss": 0.0654,
+      "step": 12850
+    },
+    {
+      "epoch": 117.9,
+      "learning_rate": 8.944954128440367e-06,
+      "loss": 0.0763,
+      "step": 12851
+    },
+    {
+      "epoch": 117.91,
+      "learning_rate": 8.906727828746178e-06,
+      "loss": 0.0757,
+      "step": 12852
+    },
+    {
+      "epoch": 117.92,
+      "learning_rate": 8.868501529051987e-06,
+      "loss": 0.1096,
+      "step": 12853
+    },
+    {
+      "epoch": 117.93,
+      "learning_rate": 8.830275229357798e-06,
+      "loss": 0.1338,
+      "step": 12854
+    },
+    {
+      "epoch": 117.94,
+      "learning_rate": 8.79204892966361e-06,
+      "loss": 0.1204,
+      "step": 12855
+    },
+    {
+      "epoch": 117.94,
+      "learning_rate": 8.753822629969419e-06,
+      "loss": 0.0541,
+      "step": 12856
+    },
+    {
+      "epoch": 117.95,
+      "learning_rate": 8.71559633027523e-06,
+      "loss": 0.0732,
+      "step": 12857
+    },
+    {
+      "epoch": 117.96,
+      "learning_rate": 8.67737003058104e-06,
+      "loss": 0.0873,
+      "step": 12858
+    },
+    {
+      "epoch": 117.97,
+      "learning_rate": 8.63914373088685e-06,
+      "loss": 0.0567,
+      "step": 12859
+    },
+    {
+      "epoch": 117.98,
+      "learning_rate": 8.600917431192661e-06,
+      "loss": 0.0665,
+      "step": 12860
+    },
+    {
+      "epoch": 117.99,
+      "learning_rate": 8.56269113149847e-06,
+      "loss": 0.0373,
+      "step": 12861
+    },
+    {
+      "epoch": 118.0,
+      "learning_rate": 8.52446483180428e-06,
+      "loss": 0.0514,
+      "step": 12862
+    },
+    {
+      "epoch": 118.01,
+      "learning_rate": 8.486238532110093e-06,
+      "loss": 0.1293,
+      "step": 12863
+    },
+    {
+      "epoch": 118.02,
+      "learning_rate": 8.448012232415902e-06,
+      "loss": 0.0985,
+      "step": 12864
+    },
+    {
+      "epoch": 118.03,
+      "learning_rate": 8.409785932721712e-06,
+      "loss": 0.0767,
+      "step": 12865
+    },
+    {
+      "epoch": 118.04,
+      "learning_rate": 8.371559633027524e-06,
+      "loss": 0.0528,
+      "step": 12866
+    },
+    {
+      "epoch": 118.05,
+      "learning_rate": 8.333333333333334e-06,
+      "loss": 0.0939,
+      "step": 12867
+    },
+    {
+      "epoch": 118.06,
+      "learning_rate": 8.295107033639143e-06,
+      "loss": 0.1076,
+      "step": 12868
+    },
+    {
+      "epoch": 118.06,
+      "learning_rate": 8.256880733944954e-06,
+      "loss": 0.0867,
+      "step": 12869
+    },
+    {
+      "epoch": 118.07,
+      "learning_rate": 8.218654434250765e-06,
+      "loss": 0.0967,
+      "step": 12870
+    },
+    {
+      "epoch": 118.08,
+      "learning_rate": 8.180428134556575e-06,
+      "loss": 0.1486,
+      "step": 12871
+    },
+    {
+      "epoch": 118.09,
+      "learning_rate": 8.142201834862386e-06,
+      "loss": 0.099,
+      "step": 12872
+    },
+    {
+      "epoch": 118.1,
+      "learning_rate": 8.103975535168195e-06,
+      "loss": 0.0876,
+      "step": 12873
+    },
+    {
+      "epoch": 118.11,
+      "learning_rate": 8.065749235474006e-06,
+      "loss": 0.0592,
+      "step": 12874
+    },
+    {
+      "epoch": 118.12,
+      "learning_rate": 8.027522935779817e-06,
+      "loss": 0.0648,
+      "step": 12875
+    },
+    {
+      "epoch": 118.13,
+      "learning_rate": 7.989296636085627e-06,
+      "loss": 0.0874,
+      "step": 12876
+    },
+    {
+      "epoch": 118.14,
+      "learning_rate": 7.951070336391438e-06,
+      "loss": 0.0443,
+      "step": 12877
+    },
+    {
+      "epoch": 118.15,
+      "learning_rate": 7.912844036697249e-06,
+      "loss": 0.1308,
+      "step": 12878
+    },
+    {
+      "epoch": 118.16,
+      "learning_rate": 7.874617737003058e-06,
+      "loss": 0.1354,
+      "step": 12879
+    },
+    {
+      "epoch": 118.17,
+      "learning_rate": 7.836391437308867e-06,
+      "loss": 0.0897,
+      "step": 12880
+    },
+    {
+      "epoch": 118.17,
+      "learning_rate": 7.79816513761468e-06,
+      "loss": 0.0802,
+      "step": 12881
+    },
+    {
+      "epoch": 118.18,
+      "learning_rate": 7.75993883792049e-06,
+      "loss": 0.0882,
+      "step": 12882
+    },
+    {
+      "epoch": 118.19,
+      "learning_rate": 7.7217125382263e-06,
+      "loss": 0.0885,
+      "step": 12883
+    },
+    {
+      "epoch": 118.2,
+      "learning_rate": 7.68348623853211e-06,
+      "loss": 0.0811,
+      "step": 12884
+    },
+    {
+      "epoch": 118.21,
+      "learning_rate": 7.645259938837921e-06,
+      "loss": 0.0725,
+      "step": 12885
+    },
+    {
+      "epoch": 118.22,
+      "learning_rate": 7.607033639143731e-06,
+      "loss": 0.0995,
+      "step": 12886
+    },
+    {
+      "epoch": 118.23,
+      "learning_rate": 7.5688073394495415e-06,
+      "loss": 0.0844,
+      "step": 12887
+    },
+    {
+      "epoch": 118.24,
+      "learning_rate": 7.530581039755352e-06,
+      "loss": 0.0749,
+      "step": 12888
+    },
+    {
+      "epoch": 118.25,
+      "learning_rate": 7.492354740061163e-06,
+      "loss": 0.0189,
+      "step": 12889
+    },
+    {
+      "epoch": 118.26,
+      "learning_rate": 7.454128440366973e-06,
+      "loss": 0.1206,
+      "step": 12890
+    },
+    {
+      "epoch": 118.27,
+      "learning_rate": 7.415902140672783e-06,
+      "loss": 0.1148,
+      "step": 12891
+    },
+    {
+      "epoch": 118.28,
+      "learning_rate": 7.377675840978593e-06,
+      "loss": 0.0663,
+      "step": 12892
+    },
+    {
+      "epoch": 118.28,
+      "learning_rate": 7.3394495412844045e-06,
+      "loss": 0.1026,
+      "step": 12893
+    },
+    {
+      "epoch": 118.29,
+      "learning_rate": 7.301223241590214e-06,
+      "loss": 0.1125,
+      "step": 12894
+    },
+    {
+      "epoch": 118.3,
+      "learning_rate": 7.262996941896025e-06,
+      "loss": 0.0863,
+      "step": 12895
+    },
+    {
+      "epoch": 118.31,
+      "learning_rate": 7.224770642201835e-06,
+      "loss": 0.1192,
+      "step": 12896
+    },
+    {
+      "epoch": 118.32,
+      "learning_rate": 7.186544342507645e-06,
+      "loss": 0.0927,
+      "step": 12897
+    },
+    {
+      "epoch": 118.33,
+      "learning_rate": 7.1483180428134555e-06,
+      "loss": 0.0552,
+      "step": 12898
+    },
+    {
+      "epoch": 118.34,
+      "learning_rate": 7.110091743119267e-06,
+      "loss": 0.074,
+      "step": 12899
+    },
+    {
+      "epoch": 118.35,
+      "learning_rate": 7.071865443425077e-06,
+      "loss": 0.0815,
+      "step": 12900
+    },
+    {
+      "epoch": 118.36,
+      "learning_rate": 7.033639143730887e-06,
+      "loss": 0.1183,
+      "step": 12901
+    },
+    {
+      "epoch": 118.37,
+      "learning_rate": 6.995412844036698e-06,
+      "loss": 0.0769,
+      "step": 12902
+    },
+    {
+      "epoch": 118.38,
+      "learning_rate": 6.9571865443425075e-06,
+      "loss": 0.0865,
+      "step": 12903
+    },
+    {
+      "epoch": 118.39,
+      "learning_rate": 6.9189602446483185e-06,
+      "loss": 0.104,
+      "step": 12904
+    },
+    {
+      "epoch": 118.39,
+      "learning_rate": 6.880733944954129e-06,
+      "loss": 0.0679,
+      "step": 12905
+    },
+    {
+      "epoch": 118.4,
+      "learning_rate": 6.842507645259939e-06,
+      "loss": 0.0438,
+      "step": 12906
+    },
+    {
+      "epoch": 118.41,
+      "learning_rate": 6.804281345565749e-06,
+      "loss": 0.0714,
+      "step": 12907
+    },
+    {
+      "epoch": 118.42,
+      "learning_rate": 6.76605504587156e-06,
+      "loss": 0.0561,
+      "step": 12908
+    },
+    {
+      "epoch": 118.43,
+      "learning_rate": 6.72782874617737e-06,
+      "loss": 0.1088,
+      "step": 12909
+    },
+    {
+      "epoch": 118.44,
+      "learning_rate": 6.689602446483181e-06,
+      "loss": 0.0463,
+      "step": 12910
+    },
+    {
+      "epoch": 118.45,
+      "learning_rate": 6.651376146788992e-06,
+      "loss": 0.045,
+      "step": 12911
+    },
+    {
+      "epoch": 118.46,
+      "learning_rate": 6.613149847094801e-06,
+      "loss": 0.0875,
+      "step": 12912
+    },
+    {
+      "epoch": 118.47,
+      "learning_rate": 6.574923547400612e-06,
+      "loss": 0.0737,
+      "step": 12913
+    },
+    {
+      "epoch": 118.48,
+      "learning_rate": 6.536697247706422e-06,
+      "loss": 0.0853,
+      "step": 12914
+    },
+    {
+      "epoch": 118.49,
+      "learning_rate": 6.498470948012233e-06,
+      "loss": 0.0612,
+      "step": 12915
+    },
+    {
+      "epoch": 118.5,
+      "learning_rate": 6.460244648318043e-06,
+      "loss": 0.0831,
+      "step": 12916
+    },
+    {
+      "epoch": 118.5,
+      "learning_rate": 6.422018348623854e-06,
+      "loss": 0.12,
+      "step": 12917
+    },
+    {
+      "epoch": 118.51,
+      "learning_rate": 6.383792048929663e-06,
+      "loss": 0.0835,
+      "step": 12918
+    },
+    {
+      "epoch": 118.52,
+      "learning_rate": 6.345565749235474e-06,
+      "loss": 0.1016,
+      "step": 12919
+    },
+    {
+      "epoch": 118.53,
+      "learning_rate": 6.3073394495412846e-06,
+      "loss": 0.0721,
+      "step": 12920
+    },
+    {
+      "epoch": 118.54,
+      "learning_rate": 6.269113149847095e-06,
+      "loss": 0.0819,
+      "step": 12921
+    },
+    {
+      "epoch": 118.55,
+      "learning_rate": 6.230886850152905e-06,
+      "loss": 0.0994,
+      "step": 12922
+    },
+    {
+      "epoch": 118.56,
+      "learning_rate": 6.192660550458716e-06,
+      "loss": 0.0755,
+      "step": 12923
+    },
+    {
+      "epoch": 118.57,
+      "learning_rate": 6.154434250764526e-06,
+      "loss": 0.0772,
+      "step": 12924
+    },
+    {
+      "epoch": 118.58,
+      "learning_rate": 6.1162079510703365e-06,
+      "loss": 0.1298,
+      "step": 12925
+    },
+    {
+      "epoch": 118.59,
+      "learning_rate": 6.0779816513761475e-06,
+      "loss": 0.0648,
+      "step": 12926
+    },
+    {
+      "epoch": 118.6,
+      "learning_rate": 6.039755351681957e-06,
+      "loss": 0.064,
+      "step": 12927
+    },
+    {
+      "epoch": 118.61,
+      "learning_rate": 6.001529051987768e-06,
+      "loss": 0.0649,
+      "step": 12928
+    },
+    {
+      "epoch": 118.61,
+      "learning_rate": 5.963302752293578e-06,
+      "loss": 0.1085,
+      "step": 12929
+    },
+    {
+      "epoch": 118.62,
+      "learning_rate": 5.925076452599388e-06,
+      "loss": 0.1195,
+      "step": 12930
+    },
+    {
+      "epoch": 118.63,
+      "learning_rate": 5.886850152905199e-06,
+      "loss": 0.109,
+      "step": 12931
+    },
+    {
+      "epoch": 118.64,
+      "learning_rate": 5.84862385321101e-06,
+      "loss": 0.0717,
+      "step": 12932
+    },
+    {
+      "epoch": 118.65,
+      "learning_rate": 5.810397553516819e-06,
+      "loss": 0.0506,
+      "step": 12933
+    },
+    {
+      "epoch": 118.66,
+      "learning_rate": 5.77217125382263e-06,
+      "loss": 0.0644,
+      "step": 12934
+    },
+    {
+      "epoch": 118.67,
+      "learning_rate": 5.733944954128441e-06,
+      "loss": 0.0892,
+      "step": 12935
+    },
+    {
+      "epoch": 118.68,
+      "learning_rate": 5.6957186544342506e-06,
+      "loss": 0.0866,
+      "step": 12936
+    },
+    {
+      "epoch": 118.69,
+      "learning_rate": 5.657492354740062e-06,
+      "loss": 0.0534,
+      "step": 12937
+    },
+    {
+      "epoch": 118.7,
+      "learning_rate": 5.619266055045872e-06,
+      "loss": 0.0616,
+      "step": 12938
+    },
+    {
+      "epoch": 118.71,
+      "learning_rate": 5.581039755351682e-06,
+      "loss": 0.0396,
+      "step": 12939
+    },
+    {
+      "epoch": 118.72,
+      "learning_rate": 5.542813455657492e-06,
+      "loss": 0.1091,
+      "step": 12940
+    },
+    {
+      "epoch": 118.72,
+      "learning_rate": 5.504587155963303e-06,
+      "loss": 0.0332,
+      "step": 12941
+    },
+    {
+      "epoch": 118.73,
+      "learning_rate": 5.466360856269113e-06,
+      "loss": 0.0391,
+      "step": 12942
+    },
+    {
+      "epoch": 118.74,
+      "learning_rate": 5.428134556574924e-06,
+      "loss": 0.0654,
+      "step": 12943
+    },
+    {
+      "epoch": 118.75,
+      "learning_rate": 5.389908256880734e-06,
+      "loss": 0.1624,
+      "step": 12944
+    },
+    {
+      "epoch": 118.76,
+      "learning_rate": 5.351681957186544e-06,
+      "loss": 0.1525,
+      "step": 12945
+    },
+    {
+      "epoch": 118.77,
+      "learning_rate": 5.313455657492355e-06,
+      "loss": 0.0871,
+      "step": 12946
+    },
+    {
+      "epoch": 118.78,
+      "learning_rate": 5.2752293577981655e-06,
+      "loss": 0.0983,
+      "step": 12947
+    },
+    {
+      "epoch": 118.79,
+      "learning_rate": 5.237003058103976e-06,
+      "loss": 0.1237,
+      "step": 12948
+    },
+    {
+      "epoch": 118.8,
+      "learning_rate": 5.198776758409786e-06,
+      "loss": 0.0913,
+      "step": 12949
+    },
+    {
+      "epoch": 118.81,
+      "learning_rate": 5.160550458715597e-06,
+      "loss": 0.0492,
+      "step": 12950
+    },
+    {
+      "epoch": 118.82,
+      "learning_rate": 5.122324159021406e-06,
+      "loss": 0.0922,
+      "step": 12951
+    },
+    {
+      "epoch": 118.83,
+      "learning_rate": 5.084097859327217e-06,
+      "loss": 0.0679,
+      "step": 12952
+    },
+    {
+      "epoch": 118.83,
+      "learning_rate": 5.045871559633028e-06,
+      "loss": 0.0952,
+      "step": 12953
+    },
+    {
+      "epoch": 118.84,
+      "learning_rate": 5.007645259938838e-06,
+      "loss": 0.0671,
+      "step": 12954
+    },
+    {
+      "epoch": 118.85,
+      "learning_rate": 4.969418960244648e-06,
+      "loss": 0.1037,
+      "step": 12955
+    },
+    {
+      "epoch": 118.86,
+      "learning_rate": 4.931192660550459e-06,
+      "loss": 0.0687,
+      "step": 12956
+    },
+    {
+      "epoch": 118.87,
+      "learning_rate": 4.8929663608562685e-06,
+      "loss": 0.0276,
+      "step": 12957
+    },
+    {
+      "epoch": 118.88,
+      "learning_rate": 4.8547400611620796e-06,
+      "loss": 0.1105,
+      "step": 12958
+    },
+    {
+      "epoch": 118.89,
+      "learning_rate": 4.816513761467891e-06,
+      "loss": 0.078,
+      "step": 12959
+    },
+    {
+      "epoch": 118.9,
+      "learning_rate": 4.7782874617737e-06,
+      "loss": 0.0922,
+      "step": 12960
+    },
+    {
+      "epoch": 118.91,
+      "learning_rate": 4.740061162079511e-06,
+      "loss": 0.0654,
+      "step": 12961
+    },
+    {
+      "epoch": 118.92,
+      "learning_rate": 4.701834862385321e-06,
+      "loss": 0.0897,
+      "step": 12962
+    },
+    {
+      "epoch": 118.93,
+      "learning_rate": 4.6636085626911315e-06,
+      "loss": 0.0623,
+      "step": 12963
+    },
+    {
+      "epoch": 118.94,
+      "learning_rate": 4.625382262996942e-06,
+      "loss": 0.045,
+      "step": 12964
+    },
+    {
+      "epoch": 118.94,
+      "learning_rate": 4.587155963302753e-06,
+      "loss": 0.087,
+      "step": 12965
+    },
+    {
+      "epoch": 118.95,
+      "learning_rate": 4.548929663608562e-06,
+      "loss": 0.0729,
+      "step": 12966
+    },
+    {
+      "epoch": 118.96,
+      "learning_rate": 4.510703363914373e-06,
+      "loss": 0.0956,
+      "step": 12967
+    },
+    {
+      "epoch": 118.97,
+      "learning_rate": 4.4724770642201834e-06,
+      "loss": 0.0518,
+      "step": 12968
+    },
+    {
+      "epoch": 118.98,
+      "learning_rate": 4.434250764525994e-06,
+      "loss": 0.0556,
+      "step": 12969
+    },
+    {
+      "epoch": 118.99,
+      "learning_rate": 4.396024464831805e-06,
+      "loss": 0.0475,
+      "step": 12970
+    },
+    {
+      "epoch": 119.0,
+      "learning_rate": 4.357798165137615e-06,
+      "loss": 0.0238,
+      "step": 12971
+    },
+    {
+      "epoch": 119.01,
+      "learning_rate": 4.319571865443425e-06,
+      "loss": 0.1304,
+      "step": 12972
+    },
+    {
+      "epoch": 119.02,
+      "learning_rate": 4.281345565749235e-06,
+      "loss": 0.0814,
+      "step": 12973
+    },
+    {
+      "epoch": 119.03,
+      "learning_rate": 4.2431192660550464e-06,
+      "loss": 0.1046,
+      "step": 12974
+    },
+    {
+      "epoch": 119.04,
+      "learning_rate": 4.204892966360856e-06,
+      "loss": 0.061,
+      "step": 12975
+    },
+    {
+      "epoch": 119.05,
+      "learning_rate": 4.166666666666667e-06,
+      "loss": 0.0438,
+      "step": 12976
+    },
+    {
+      "epoch": 119.06,
+      "learning_rate": 4.128440366972477e-06,
+      "loss": 0.0931,
+      "step": 12977
+    },
+    {
+      "epoch": 119.06,
+      "learning_rate": 4.090214067278287e-06,
+      "loss": 0.0853,
+      "step": 12978
+    },
+    {
+      "epoch": 119.07,
+      "learning_rate": 4.0519877675840975e-06,
+      "loss": 0.0333,
+      "step": 12979
+    },
+    {
+      "epoch": 119.08,
+      "learning_rate": 4.013761467889909e-06,
+      "loss": 0.0457,
+      "step": 12980
+    },
+    {
+      "epoch": 119.09,
+      "learning_rate": 3.975535168195719e-06,
+      "loss": 0.0926,
+      "step": 12981
+    },
+    {
+      "epoch": 119.1,
+      "learning_rate": 3.937308868501529e-06,
+      "loss": 0.125,
+      "step": 12982
+    },
+    {
+      "epoch": 119.11,
+      "learning_rate": 3.89908256880734e-06,
+      "loss": 0.092,
+      "step": 12983
+    },
+    {
+      "epoch": 119.12,
+      "learning_rate": 3.86085626911315e-06,
+      "loss": 0.0612,
+      "step": 12984
+    },
+    {
+      "epoch": 119.13,
+      "learning_rate": 3.8226299694189605e-06,
+      "loss": 0.1016,
+      "step": 12985
+    },
+    {
+      "epoch": 119.14,
+      "learning_rate": 3.7844036697247707e-06,
+      "loss": 0.1091,
+      "step": 12986
+    },
+    {
+      "epoch": 119.15,
+      "learning_rate": 3.7461773700305814e-06,
+      "loss": 0.0593,
+      "step": 12987
+    },
+    {
+      "epoch": 119.16,
+      "learning_rate": 3.7079510703363916e-06,
+      "loss": 0.1368,
+      "step": 12988
+    },
+    {
+      "epoch": 119.17,
+      "learning_rate": 3.6697247706422022e-06,
+      "loss": 0.0701,
+      "step": 12989
+    },
+    {
+      "epoch": 119.17,
+      "learning_rate": 3.6314984709480124e-06,
+      "loss": 0.0653,
+      "step": 12990
+    },
+    {
+      "epoch": 119.18,
+      "learning_rate": 3.5932721712538227e-06,
+      "loss": 0.0637,
+      "step": 12991
+    },
+    {
+      "epoch": 119.19,
+      "learning_rate": 3.5550458715596333e-06,
+      "loss": 0.1093,
+      "step": 12992
+    },
+    {
+      "epoch": 119.2,
+      "learning_rate": 3.5168195718654435e-06,
+      "loss": 0.0902,
+      "step": 12993
+    },
+    {
+      "epoch": 119.21,
+      "learning_rate": 3.4785932721712537e-06,
+      "loss": 0.0501,
+      "step": 12994
+    },
+    {
+      "epoch": 119.22,
+      "learning_rate": 3.4403669724770644e-06,
+      "loss": 0.0313,
+      "step": 12995
+    },
+    {
+      "epoch": 119.23,
+      "learning_rate": 3.4021406727828746e-06,
+      "loss": 0.0422,
+      "step": 12996
+    },
+    {
+      "epoch": 119.24,
+      "learning_rate": 3.363914373088685e-06,
+      "loss": 0.0541,
+      "step": 12997
+    },
+    {
+      "epoch": 119.25,
+      "learning_rate": 3.325688073394496e-06,
+      "loss": 0.0205,
+      "step": 12998
+    },
+    {
+      "epoch": 119.26,
+      "learning_rate": 3.287461773700306e-06,
+      "loss": 0.1206,
+      "step": 12999
+    },
+    {
+      "epoch": 119.27,
+      "learning_rate": 3.2492354740061163e-06,
+      "loss": 0.1122,
+      "step": 13000
+    },
+    {
+      "epoch": 119.27,
+      "eval_cer": 0.1316811346496949,
+      "eval_loss": 0.7534573078155518,
+      "eval_runtime": 87.8463,
+      "eval_samples_per_second": 18.749,
+      "eval_steps_per_second": 2.345,
+      "eval_wer": 0.46512102951690326,
+      "step": 13000
+    },
+    {
+      "epoch": 119.28,
+      "learning_rate": 3.211009174311927e-06,
+      "loss": 0.0854,
+      "step": 13001
+    },
+    {
+      "epoch": 119.28,
+      "learning_rate": 3.172782874617737e-06,
+      "loss": 0.0569,
+      "step": 13002
+    },
+    {
+      "epoch": 119.29,
+      "learning_rate": 3.1345565749235474e-06,
+      "loss": 0.0971,
+      "step": 13003
+    },
+    {
+      "epoch": 119.3,
+      "learning_rate": 3.096330275229358e-06,
+      "loss": 0.0785,
+      "step": 13004
+    },
+    {
+      "epoch": 119.31,
+      "learning_rate": 3.0581039755351682e-06,
+      "loss": 0.0469,
+      "step": 13005
+    },
+    {
+      "epoch": 119.32,
+      "learning_rate": 3.0198776758409785e-06,
+      "loss": 0.0882,
+      "step": 13006
+    },
+    {
+      "epoch": 119.33,
+      "learning_rate": 2.981651376146789e-06,
+      "loss": 0.0706,
+      "step": 13007
+    },
+    {
+      "epoch": 119.34,
+      "learning_rate": 2.9434250764525993e-06,
+      "loss": 0.0895,
+      "step": 13008
+    },
+    {
+      "epoch": 119.35,
+      "learning_rate": 2.9051987767584095e-06,
+      "loss": 0.062,
+      "step": 13009
+    },
+    {
+      "epoch": 119.36,
+      "learning_rate": 2.8669724770642206e-06,
+      "loss": 0.0517,
+      "step": 13010
+    },
+    {
+      "epoch": 119.37,
+      "learning_rate": 2.828746177370031e-06,
+      "loss": 0.1047,
+      "step": 13011
+    },
+    {
+      "epoch": 119.38,
+      "learning_rate": 2.790519877675841e-06,
+      "loss": 0.0842,
+      "step": 13012
+    },
+    {
+      "epoch": 119.39,
+      "learning_rate": 2.7522935779816517e-06,
+      "loss": 0.0765,
+      "step": 13013
+    },
+    {
+      "epoch": 119.39,
+      "learning_rate": 2.714067278287462e-06,
+      "loss": 0.0469,
+      "step": 13014
+    },
+    {
+      "epoch": 119.4,
+      "learning_rate": 2.675840978593272e-06,
+      "loss": 0.0997,
+      "step": 13015
+    },
+    {
+      "epoch": 119.41,
+      "learning_rate": 2.6376146788990827e-06,
+      "loss": 0.0983,
+      "step": 13016
+    },
+    {
+      "epoch": 119.42,
+      "learning_rate": 2.599388379204893e-06,
+      "loss": 0.0993,
+      "step": 13017
+    },
+    {
+      "epoch": 119.43,
+      "learning_rate": 2.561162079510703e-06,
+      "loss": 0.0637,
+      "step": 13018
+    },
+    {
+      "epoch": 119.44,
+      "learning_rate": 2.522935779816514e-06,
+      "loss": 0.0583,
+      "step": 13019
+    },
+    {
+      "epoch": 119.45,
+      "learning_rate": 2.484709480122324e-06,
+      "loss": 0.0322,
+      "step": 13020
+    },
+    {
+      "epoch": 119.46,
+      "learning_rate": 2.4464831804281343e-06,
+      "loss": 0.0873,
+      "step": 13021
+    },
+    {
+      "epoch": 119.47,
+      "learning_rate": 2.4082568807339453e-06,
+      "loss": 0.0464,
+      "step": 13022
+    },
+    {
+      "epoch": 119.48,
+      "learning_rate": 2.3700305810397555e-06,
+      "loss": 0.0486,
+      "step": 13023
+    },
+    {
+      "epoch": 119.49,
+      "learning_rate": 2.3318042813455658e-06,
+      "loss": 0.0486,
+      "step": 13024
+    },
+    {
+      "epoch": 119.5,
+      "learning_rate": 2.2935779816513764e-06,
+      "loss": 0.0183,
+      "step": 13025
+    },
+    {
+      "epoch": 119.5,
+      "learning_rate": 2.2553516819571866e-06,
+      "loss": 0.1524,
+      "step": 13026
+    },
+    {
+      "epoch": 119.51,
+      "learning_rate": 2.217125382262997e-06,
+      "loss": 0.0568,
+      "step": 13027
+    },
+    {
+      "epoch": 119.52,
+      "learning_rate": 2.1788990825688075e-06,
+      "loss": 0.0954,
+      "step": 13028
+    },
+    {
+      "epoch": 119.53,
+      "learning_rate": 2.1406727828746177e-06,
+      "loss": 0.1196,
+      "step": 13029
+    },
+    {
+      "epoch": 119.54,
+      "learning_rate": 2.102446483180428e-06,
+      "loss": 0.0774,
+      "step": 13030
+    },
+    {
+      "epoch": 119.55,
+      "learning_rate": 2.0642201834862385e-06,
+      "loss": 0.0678,
+      "step": 13031
+    },
+    {
+      "epoch": 119.56,
+      "learning_rate": 2.0259938837920488e-06,
+      "loss": 0.0432,
+      "step": 13032
+    },
+    {
+      "epoch": 119.57,
+      "learning_rate": 1.9877675840978594e-06,
+      "loss": 0.0622,
+      "step": 13033
+    },
+    {
+      "epoch": 119.58,
+      "learning_rate": 1.94954128440367e-06,
+      "loss": 0.0797,
+      "step": 13034
+    },
+    {
+      "epoch": 119.59,
+      "learning_rate": 1.9113149847094803e-06,
+      "loss": 0.1085,
+      "step": 13035
+    },
+    {
+      "epoch": 119.6,
+      "learning_rate": 1.8730886850152907e-06,
+      "loss": 0.1096,
+      "step": 13036
+    },
+    {
+      "epoch": 119.61,
+      "learning_rate": 1.8348623853211011e-06,
+      "loss": 0.1229,
+      "step": 13037
+    },
+    {
+      "epoch": 119.61,
+      "learning_rate": 1.7966360856269113e-06,
+      "loss": 0.087,
+      "step": 13038
+    },
+    {
+      "epoch": 119.62,
+      "learning_rate": 1.7584097859327218e-06,
+      "loss": 0.0716,
+      "step": 13039
+    },
+    {
+      "epoch": 119.63,
+      "learning_rate": 1.7201834862385322e-06,
+      "loss": 0.0942,
+      "step": 13040
+    },
+    {
+      "epoch": 119.64,
+      "learning_rate": 1.6819571865443424e-06,
+      "loss": 0.0686,
+      "step": 13041
+    },
+    {
+      "epoch": 119.65,
+      "learning_rate": 1.643730886850153e-06,
+      "loss": 0.1431,
+      "step": 13042
+    },
+    {
+      "epoch": 119.66,
+      "learning_rate": 1.6055045871559635e-06,
+      "loss": 0.1038,
+      "step": 13043
+    },
+    {
+      "epoch": 119.67,
+      "learning_rate": 1.5672782874617737e-06,
+      "loss": 0.0516,
+      "step": 13044
+    },
+    {
+      "epoch": 119.68,
+      "learning_rate": 1.5290519877675841e-06,
+      "loss": 0.0647,
+      "step": 13045
+    },
+    {
+      "epoch": 119.69,
+      "learning_rate": 1.4908256880733945e-06,
+      "loss": 0.0996,
+      "step": 13046
+    },
+    {
+      "epoch": 119.7,
+      "learning_rate": 1.4525993883792048e-06,
+      "loss": 0.0659,
+      "step": 13047
+    },
+    {
+      "epoch": 119.71,
+      "learning_rate": 1.4143730886850154e-06,
+      "loss": 0.0431,
+      "step": 13048
+    },
+    {
+      "epoch": 119.72,
+      "learning_rate": 1.3761467889908258e-06,
+      "loss": 0.0359,
+      "step": 13049
+    },
+    {
+      "epoch": 119.72,
+      "learning_rate": 1.337920489296636e-06,
+      "loss": 0.0576,
+      "step": 13050
+    },
+    {
+      "epoch": 119.73,
+      "learning_rate": 1.2996941896024465e-06,
+      "loss": 0.0453,
+      "step": 13051
+    },
+    {
+      "epoch": 119.74,
+      "learning_rate": 1.261467889908257e-06,
+      "loss": 0.0324,
+      "step": 13052
+    },
+    {
+      "epoch": 119.75,
+      "learning_rate": 1.2232415902140671e-06,
+      "loss": 0.1434,
+      "step": 13053
+    },
+    {
+      "epoch": 119.76,
+      "learning_rate": 1.1850152905198778e-06,
+      "loss": 0.1523,
+      "step": 13054
+    },
+    {
+      "epoch": 119.77,
+      "learning_rate": 1.1467889908256882e-06,
+      "loss": 0.0999,
+      "step": 13055
+    },
+    {
+      "epoch": 119.78,
+      "learning_rate": 1.1085626911314984e-06,
+      "loss": 0.0943,
+      "step": 13056
+    },
+    {
+      "epoch": 119.79,
+      "learning_rate": 1.0703363914373088e-06,
+      "loss": 0.0894,
+      "step": 13057
+    },
+    {
+      "epoch": 119.8,
+      "learning_rate": 1.0321100917431193e-06,
+      "loss": 0.0783,
+      "step": 13058
+    },
+    {
+      "epoch": 119.81,
+      "learning_rate": 9.938837920489297e-07,
+      "loss": 0.0994,
+      "step": 13059
+    },
+    {
+      "epoch": 119.82,
+      "learning_rate": 9.556574923547401e-07,
+      "loss": 0.0576,
+      "step": 13060
+    },
+    {
+      "epoch": 119.83,
+      "learning_rate": 9.174311926605506e-07,
+      "loss": 0.0705,
+      "step": 13061
+    },
+    {
+      "epoch": 119.83,
+      "learning_rate": 8.792048929663609e-07,
+      "loss": 0.0641,
+      "step": 13062
+    },
+    {
+      "epoch": 119.84,
+      "learning_rate": 8.409785932721712e-07,
+      "loss": 0.1253,
+      "step": 13063
+    },
+    {
+      "epoch": 119.85,
+      "learning_rate": 8.027522935779817e-07,
+      "loss": 0.0899,
+      "step": 13064
+    },
+    {
+      "epoch": 119.86,
+      "learning_rate": 7.645259938837921e-07,
+      "loss": 0.077,
+      "step": 13065
+    },
+    {
+      "epoch": 119.87,
+      "learning_rate": 7.262996941896024e-07,
+      "loss": 0.032,
+      "step": 13066
+    },
+    {
+      "epoch": 119.88,
+      "learning_rate": 6.880733944954129e-07,
+      "loss": 0.0462,
+      "step": 13067
+    },
+    {
+      "epoch": 119.89,
+      "learning_rate": 6.498470948012232e-07,
+      "loss": 0.0897,
+      "step": 13068
+    },
+    {
+      "epoch": 119.9,
+      "learning_rate": 6.116207951070336e-07,
+      "loss": 0.0622,
+      "step": 13069
+    },
+    {
+      "epoch": 119.91,
+      "learning_rate": 5.733944954128441e-07,
+      "loss": 0.1148,
+      "step": 13070
+    },
+    {
+      "epoch": 119.92,
+      "learning_rate": 5.351681957186544e-07,
+      "loss": 0.0638,
+      "step": 13071
+    },
+    {
+      "epoch": 119.93,
+      "learning_rate": 4.969418960244648e-07,
+      "loss": 0.0635,
+      "step": 13072
+    },
+    {
+      "epoch": 119.94,
+      "learning_rate": 4.587155963302753e-07,
+      "loss": 0.0813,
+      "step": 13073
+    },
+    {
+      "epoch": 119.94,
+      "learning_rate": 4.204892966360856e-07,
+      "loss": 0.0825,
+      "step": 13074
+    },
+    {
+      "epoch": 119.95,
+      "learning_rate": 3.8226299694189603e-07,
+      "loss": 0.0552,
+      "step": 13075
+    },
+    {
+      "epoch": 119.96,
+      "learning_rate": 3.4403669724770646e-07,
+      "loss": 0.0686,
+      "step": 13076
+    },
+    {
+      "epoch": 119.97,
+      "learning_rate": 3.058103975535168e-07,
+      "loss": 0.1507,
+      "step": 13077
+    },
+    {
+      "epoch": 119.98,
+      "learning_rate": 2.675840978593272e-07,
+      "loss": 0.039,
+      "step": 13078
+    },
+    {
+      "epoch": 119.99,
+      "learning_rate": 2.2935779816513764e-07,
+      "loss": 0.1128,
+      "step": 13079
+    },
+    {
+      "epoch": 120.0,
+      "learning_rate": 1.9113149847094802e-07,
+      "loss": 0.0626,
+      "step": 13080
+    },
+    {
+      "epoch": 120.0,
+      "step": 13080,
+      "total_flos": 1.6034049995906177e+20,
+      "train_loss": 0.020454988959786518,
+      "train_runtime": 6971.2586,
+      "train_samples_per_second": 59.869,
+      "train_steps_per_second": 1.876
     }
   ],
-  "max_steps": 545,
-  "num_train_epochs": 5,
-  "total_flos": 1.2873670788396168e+20,
+  "max_steps": 13080,
+  "num_train_epochs": 120,
+  "total_flos": 1.6034049995906177e+20,
   "trial_name": null,
   "trial_params": null
 }